The history of BS 7910 flaw interaction

Hunting Energy Services (UK) Ltd

John Sharples
Serco Technical Consultancy Services

Isabel Hadley, Henryk Pisarski 
TWI Ltd

Paper presented at ASME 2011 Pressure Vessels & Piping Division Conference – PVP 2011 – July 17-21 2011, Baltimore, Maryland, USA

Abstract

Assessment of flaws and flaw-like indications in engineering components is commonly performed using fitness for service procedures such as the BS 7910 to demonstrate margins for continued operation or to find a limiting state. An individual or multiple flaw indications may be found in a component as a result of fabrication or degradation during operation. Most fitness for service codes provide guidance for assessment of multiple flaws. This consists of simplifying the actual indications into flaws more amenable to analysis using engineering criteria.

The subject of this paper is the flaw interaction criteria used by the BS 7910 procedure. The criteria have been refined several times since the 1980 version of the BS 7910 predecessor document PD 6493. This paper will summarise the changes and provide plausible explanations to the changes supported by the experimental data and numerical analysis. A proposed way forward for the 2012 edition of BS 7910 flaw interaction and characterisation will conclude the paper. 

1. Introduction

The remaining structural integrity of an engineering component containing a crack or crack-like flaw is performed by engineering critical assessment (ECA) using fitness for service procedures such as the British Standard BS 7910 [1] or the American Society of Mechanical Engineers’ Boiler and Pressure Vessel code, Section XI [2]. The fundamental objective of these procedures is to guide the user in demonstrating continued operation / repair / replacement decisions or to find a limiting state. The procedures differ in the steps taken to achieve the objective. For the assessment of fracture British Standard BS 7910 performs fracture mechanics based assessment over several levels of complexity from the onset, whereas the ASME Boiler and Pressure Vessel code first utilises flaw acceptance tables and gives the user options in the non-mandatory appendices to perform detailed assessment for fracture. Other procedures, such as FITNET [3] or R6 [4] generally follow the full assessment route as outlined by the BS 7910.

BS 7910 has evolved from the PD 6493 [5] first published in 1980 and revised in 1991. In 1999 the British Standard BS 7910 was first published and essentially utilised most of the PD 6493:1991 procedures with updates to reflect the growing body of knowledge. A minor revision of BS 7910 was published in 2005 and the next major revision is planned for 2012.

2. The BS 7910 Procedure

The scope of the BS 7910 procedure is to guide the user in assessing the acceptability of flaws in metallic structures. The procedure can be used in two ways:

• to perform an assessment of an existing flaw and support continued operation / repair / replacement decisions
• to determine a limiting flaw size for a component subject to a given operating environment and determine the time to reach the limiting state.   

The engineering critical assessment is performed in both cases. Essential to the ECA is reliable information on possible flaw size and shape and flaw location. In this regard the non-destructive examination (NDE) is essential to the process. Guidelines on performing NDE are however not the subject of BS 7910 and can be found in BS EN 571-1, 1289, 1290, 1291, 1435, 1712, 1714, 12517 and BS EN ISO 6520-1.

The next key ingredient in performing an ECA is information about the material properties:

• tensile data to  BS EN 10002-1 and 10002-5
• fatigue data to BS ISO 12108
• fracture toughness data to BS 7448, BS EN ISO 15653 and BS EN ISO 12737
• stress corrosion data to BS EN ISO 7539

BS 7910 addresses common failure mechanisms – fracture, plastic collapse, fatigue, corrosion and creep as well as more application specific issues, such as buckling or leak before break. It is imperative to an ECA to establish the cause of cracking / material degradation which may involve review of the design parameters, operating environment, material in-service degradation, etc.

The essential steps in performing an assessment are:

• identify the flaw type
• establish the essential data for the structure (material, stresses, operating environment, etc.)
• determine the size (and shape) of the flaw and distribution if multiple flaws are present
• assess material damage mechanisms and damage rates
• determine the limiting flaw size for the final failure mode
• assess the subcritical crack growth based on the  damage mechanisms
• evaluate the consequences of failure
• perform sensitivity analyses.

The flaw is deemed acceptable if it does not grow to the limiting state during the inspection interval, including the appropriate safety factors.

The stresses to be used in the assessment are those calculated for the unflawed structure, across the entire section thickness. The actual stress distribution may be used or stresses can be linearised. Stresses are classified into primary and secondary and each sub-divided into membrane and bending components. Bending stresses are defined across the section thickness containing the flaw and not across the component.

The assessment for fracture is performed using several analysis options of increasing complexity / reduced
in-built conservatism. Failure assessment diagrams (FADs) are used as a basis for assessment. Detailed guidance can be found in BS 7910 [1]. The key steps are outlined here to set the remainder of the paper into context:

• Define stresses
• Define fracture toughness
• Determine material properties
• Characterise the flaw
• Select the failure assessment diagram
• Calculate Lr (plastic collapse axis of FAD)and Kr (fracture axis of FAD) and plot the assessment point in the diagram
• Assess the significance of the result

These steps in the assessment were set by the PD 6493:1991 and have remained largely unchanged.  Some changes are being made to the specifics of these steps as the current standard develops into the new standard (2012 revision of BS7910). This paper aims to summarise and explain the evolution of the flaw characterisation step central to which are the interaction criteria.  

3. Flaw Characterisation

The first objective of flaw characterisation is to resolve an NDE indication into a flaw more amenable to analysis. The key actions are:

• classifying the flaw as surface breaking or subsurface
• resolving the shape of the flaw
• resolving the size of the flaw

The NDE indication is bound by a containment rectangle, one side of which is typically drawn parallel to the free surface, as shown in Figure 1. The flaw is then idealised as an elliptical flaw, for the case of embedded flaws, or a semi-elliptical flaw for the case of surface-breaking flaws, which is inscribed within the containment rectangle. The ECA is then performed on this flaw. This fundamental step has not changed since BS PD 6493:1980.

The second objective of the flaw characterisation step is to resolve multiple interacting NDE indications. Here, a distinction is made between flaws on the same cross-sectional plane, termed co-planar flaws and non-coplanar flaws. Flaws are classified as surface breaking, subsurface or a combination of the two. The interaction is judged by comparing the relevant dimensions of a flaw with the flaw interaction criteria, discussed below. This is continuously being refined to reflect the growing body of knowledge on this topic.

3.1 Flaw interaction criteria
During the development of PD 6493 and BS 7910 changes have been made to the flaw interaction criteria for
co-planar surface flaws.. The criteria are summarised in Table 1. For the definition of flaw parameters see Figure 1.

Table 1: The Flaw interaction criteria of PD 6493 and BS 7910 for co-planar surface flaws

The PD 6493: 1980 based the interaction criterion on the average surface lengths of the two flaws. The 1991 version revised this criterion to the surface length of the smaller (shorter) flaw. Finally, the BS 7910 in 1999 introduced the criterion of interaction when the two flaws touch (s=0) for flaws of a low aspect ratio (i.e. flaws with depth less than half surface length).

The aim of the interaction criterion is to find a balance between the inherent pessimism involved with such simplifications and the evidence from experimental test data and, more recently,  finite element analyses.

The basis for interaction criterion used in the 1980 version of PD 6493 was a 20% increase in the stress intensity factor on the first flaw by the second flaw. According to diagrams provided, the second flaw was the larger of the two. The interaction criterion was based on linear elastic solutions. In addition, the interaction rules were the same for fracture and fatigue.

Figure 2 illustrates the amount of conservatism in the PD 6493:1980 version of the flaw interaction criterion. The figure uses the data of Bezensek and Hancock [5, 6] to show the interaction factor as a function of flaw separation. The interaction factor is defined as a ratio of the stress intensity factors for the adjacent to the remote crack tip. Both are derived from the fatigue crack growth data using the Paris law [see reference 5 for details]. The flaw separation is normalised by the average surface length in accordance with PD 6493:1980 criterion. Notable interaction is observed only when the flaw separation, s, is less than 20% of the average flaw surface length. Clearly the criterion of PD 6493: 1980 is conservative.

To reduce, what can be at times, excessive conservatism, the 1991 revision of the PD 6493 adopted the criterion based on the shorter of the two flaw surface lengths. This makes no difference to the data of Bezensek and Hancock in Figure 2, since their focus was on two identical interacting flaws. An argument can, however, be made in support of basing the interaction criterion on the shorter of the two flaws: if one flaw is smaller in size compared with the other flaw, the magnitude of interaction will also be smaller and in the limit of one flaw disappearing no interaction will be present.  This view is supported by the numerical analysis of Hasegawa et al [7] shown in Figure 3, based on the elastic stress intensity factor calculations.  

In the late 1980’s and 1990’s there was a series of fatigue crack growth tests performed on adjacent co-planar flaws [8-10].

Similar tests were also conducted by Bezensek & Hancock in the early 2000’s [5, 6] and more recently, Sharples et al [12-14]. These tests studied the fatigue crack growth rates during the growth of co-planar flaws, from the initial independent flaws, to the interacting flaws and finally into a bounding flaw, as shown in Figure 4 after [5]. The consistent outcome in all of the tests was an observation that the flaws marginally accelerate in growth towards the adjacent flaw near the free surface when the adjacent crack tips are very close, as shown in Figure 5. Using the Paris law, the stress intensity factors can be extracted from this data as shown in Figure 2.

In the late 1990’s a review of the flaw interaction for co-planar flaws was carried out within the ASME Boiler and Pressure Vessel Code, Section XI. The summary of that study can be found in a paper by Hasegawa et al [7]. The study was based on the stress intensity factors for adjacent surface and embedded flaws and concluded that flaw interaction becomes significant when the two flaws approach one another within the length comparable to the flaw depth. An example is shown in Figure 3 from reference [7]. This conclusion was in part supported by the fatigue test data obtained by one of the present authors [5] and formed the basis for the revision to the flaw combination rule in the 2003 Addenda to the Boiler code, Section XI. Similar reviews can also be found in literature.

The intent of PD 6493:1980 and 1991 was to preclude excessive interaction by characterising the actual flaws into a larger bounding flaw. This is inherently conservative as all the fatigue life, from the onset of fulfilling the flaw interaction criterion to the achievement of the bounding flaw, is neglected.  

In BS 7910: 1999 the criterion was further relaxed by allowing the adjacent flaws to touch prior to performing flaw characterisation. It may be argued that the overpredicted fatigue life, by not accounting for the interaction in the period of growth up to flaws touching, is compensated by characterising the touching flaws into a bounding flaw. Figure 6 illustrates this idea. This argument could be used in justifying the revision of the flaw interaction criterion to s=0 in 1999.

Studies in [5, 6] support the use of s=0 as the flaw interaction criterion in fatigue. There is however an issue with this criterion for assessment of fracture, but especially cleavage, as discussed later.

3.2 Other changes related to the flaw characterisation since 1980

In parallel with the changes to the flaw interaction criterion, other changes were also made as follows:

• Requirement of performing assessment of the interaction of a combined (interacted) flaw with other neighbouring flaws in the 1980 and 1991 revisions of PD 6493. This requirement is no longer in the BS 7910: 1999 and later editions.
• Use of partial safety factors is required on the dimensions of a combined (interacted) flaw, if upper bound estimates of flaw size are not used in the ECA. 
• Check on the implication of the flaw ligament recategorisation on the failure mode. For example, a re-characterised flaw from a surface breaking to a through-wall flaw may result in a local buckling failure mode that would not be predicted for the original surface flaw. This requirement was in the PD 6493:1980 and 1991 revisions and is described in Annex E of
  BS 7910:2005. However, it is applicable to fracture assessments only.
• Guidance on uninspectable regions was given in PD 6493: 1980 and 1991 and is no longer stated in BS 7910: 1999 and later revisions. The guidance required that the assumption of a flaw size equal to the size of the uninspectable region should be made.

3.3 Flaw interaction in ductile tearing

The majority of studies on flaw interaction and characterisation were based on the fatigue test data and associated stress intensity factor calculations. A study by Bezensek and Hancock [5] looked at characterisation of the co-planar flaws on the upper shelf under large amounts of ductile tearing. The study concluded that touching co-planar flaws evolve into a bounding flaw in a similar manner to that observed in fatigue, as shown in Figure 7. The load capacity of the bounding flaw will be proportional to the remaining ligament (excluding the strain hardening effects) and will be less than the capacity of the original co-planar flaws. In this context the use of the flaw interaction criterion of s=0 of BS 7910: 1999 and 2005 is conservative for ductile tearing.

3.4 Flaw interaction in cleavage 

Recent studies on flaw characterisation in cleavage by Bezensek & Hancock [11] and Sharples et al [12-14] show that characterisation of co-planar touching flaws may not always be conservative. The issue arises for conditions of limited ductility negating potential benefits of localised plasticity. Both, test data and an analysis using a weakest link fracture based approach indicate that a lower failure load will be attained at a given failure probability for touching flaws compared to a characterised flaw, as shown in Figure 8. This was confirmed by test data although the number of tests was small.

In the PVP2010-25134 paper [15] an engineering approach was presented to determine cases where the characterised flaw may not be conservative for the purpose of ECA. The approach consists of evaluating the crack tip plastic zone using Irwin’s expression and comparing with the size of the local geometric feature measured on the flaw. The approach can be further simplified and made more conservative by using the maximum extent of a flaw for comparisons with the plastic zone size instead of the more difficult to obtain local crack feature.

The maximum extent is defined as the depth for surface flaws or the larger of the two orthogonal dimensions defining the embedded elliptical flaw.  In cases where a small crack tip plastic zone is anticipated relative to the size of the flaw, two options are proposed:

1. increase the characterised flaw size
2. increase the stress intensity factor for the original characterised flaw. 

Further development of this approach is discussed in the companion paper PVP2011-57854.  

3.5 Flaw interaction for other failure mechanisms 

The flaw interaction rules of BS 7910 are sometimes used for other failure mechanisms, such as stress corrosion cracking or creep. There is a lack of data on flaw interaction under these failure mechanisms. The interaction rules developed on the basis of fatigue crack growth are typically used for such cases.

4. The way forward for the 2012 revision of BS 7910 

Several changes are planned for the 2012 revision to align BS 7910 closer with the FITNET procedure, if considered appropriate by the BSI committee. The key developments are summarised in the reference [16] by Hadley et al.

In the context of the flaw characterisation and flaw interaction, the following changes are planned:

• Provide more guidance on constructing the containment rectangle during flaw characterisation. The intent is to prescribe that the rectangle should be aligned with the pressure retaining boundary or the free surface whenever possible.
• Separate the flaw interaction criteria into the criterion for alignment of non-coplanar flaws and the criterion for combination of co-planar flaws to be consistent with the ASME Boiler code, Section XI and the FITNET.
• Change the flaw interaction criterion for
  co-planar flaws from s=0 to the half of the maximum flaw depth (i.e. s=0.5*max(a1,a2)) in agreement with the ASME Boiler code, Section XI and the FITNET.
• Provide guidance for ECA of touching flaws under cleavage where characterisation is not necessarily conservative. This will involve performing a check on crack tip plasticity and using a larger bounding flaw or an increased stress intensity factor for the bounding flaw.

5. Conclusions

A summary of the evolution of the flaw interaction criterion in the British standard BS 7910 and its predecessor PD 6493 has been given. Relatively large conservatism contained in the early versions of the interaction criterion has been reduced. Recent data however indicates that some changes are required to maintain the applicability of the flaw interaction criterion over the main failure modes and to maintain consistency with other major fitness-for-service codes. 

6. Acknowledgements

The paper is published by permission of Hunting Energy Services (UK) Ltd, Serco Technical Consulting Services and TWI Ltd.

7. References

1. BS 7910. Guidance on methods for assessing the acceptability of flaws in metallic structures (including Amendment 1). Chapter 7. British Standard Institution. London. UK. 2005
2. ASME, Boiler code, Section XI, American Society of Mechanical Engineers, Philadelphia, USA.
3. FITNET Fitness-For-Service procedure, Revision MK8, GKSS Research Centre, Geesthacht, Germany, 2008
4. R6. Assessment of the integrity of structures containing defects. Revision 4, (Latest revision, 2010), British Energy Generation Ltd. Gloucester.
5. Bezensek B. & Hancock JW., "The Re-characterisation of Complex Defects, Part I: Fatigue and Ductile tearing", Engng. Fract, Mech., 2004,71:1000-1020 
6. Bezensek B. & Hancock JW., "Brittle fracture from interacting surface breaking defects", In: PVP-Vol. 423, 2001 ASME Pressure Vessels and Piping conference , Atlanta, USA, July 2001, pp: 25-31
7. Hasegawa, K., Miyazaki, K., Kanno, S., “Interaction criteria for multiple flaws on the basis of stress intensity factors.” 2001 ASME Pressure Vessels and Piping conference, Atlanta, Ga, 22-26 July 2001
8. Soboyejo WO. Knott JF. Walsh MJ. Cropper KR. “Fatigue crack propagation of coplanar semi-elliptical cracks in pure bending.” Engng. Fract. Mech. 1990;37:323
9. Leek TH. Howard IC. “An examination of methods of assessing interacting surface cracks by comparison with experimental data.” Int J Pressure Vessels Piping. 1996;68:181-201
10. Bayley CJ. Bell R. “Experimental and numerical investigation of coplanar fatigue crack coalescence.” Int. J. Pressure Vessel Piping. 1997;74:33-37
11. Bezensek B. & Hancock JW., "The Re-characterisation of Complex Defects, Part II: Cleavage", Engng. Fract, Mech, 2004,71:1021-1040