Sunteți pe pagina 1din 13

International Journal of Pressure Vessels and Piping 77 (2000) 855867

www.elsevier.com/locate/ijpvp

Basic principles of analytical aw assessment methods


U. Zerbst a,*, R.A. Ainsworth b, K.-H. Schwalbe a
a

Institute of Materials Research, GKSS Research Centre, Max Planck Str., D-21502 Geesthacht, Germany b British Energy Generation Limited, Barnwood, Gloucester, GL4 3RS, UK Received 10 May 2000; revised 7 October 2000; accepted 9 October 2000

Abstract Analytical aw assessment methods play an important role in the industrial realisation of fracture mechanics application. In the eld of general as well as specic standards and guidelines there have been rapid developments in recent years. This paper gives a brief review of some of the more important methods, which have been developed over the last decades. Descriptions are given of various design curves and failure assessment diagrams, net section yielding and EPRI type solutions, the reference stress method and approaches derived from it, the ETM and the recently developed European aw assessment method SINTAP. The discussion of these approaches is restricted to the basic principles of each method. q 2001 Elsevier Science Ltd. All rights reserved.
Keywords: Crack driving force; Failure assessment diagram; Analytical aw assessment

1. Introduction In the common design philosophy, the applied stresses are compared with a limit stress such as the yield strength of the material. As long as the latter exceeds the applied stresses, the component is regarded as safe. The implicit background assumption is that the component is defect-free. If a real or assumed crack or crack-like aw affects the load carrying capacity, fracture mechanics has to be applied. Then the comparison between the applied and the material side has to be carried out on the basis of crack tip parameters such as the linear elastic stress intensity factor, K, the J integral or the crack tip opening displacement (CTOD). As a result, the fracture behaviour of the component can be predicted in terms of a critical applied load or a critical crack size. Standardised solutions for the crack tip parameters are available for test specimens which are used for measuring the material's resistance to fracture. As long as the deformation behaviour of the structural component is linear elastic, then the relevant applied parameter in the component is K. Comprehensive compendia of K factor solutions exist in handbook format and as computer programs. The linear elastic handbook solutions are usually approximations of nite element solutions, which have been generated for a range of component and crack dimensions. Variations in the
* Corresponding author. Tel.: 149-4152-872-611; fax: 149-4152-872625. E-mail address: uwe.zerbst@gkss.de (U. Zerbst).

loading geometry can be considered, for example, by applying the weight function method. If the component behaves in an elasticplastic manner, the situation is much more complex because the crack tip loading is additionally inuenced by the deformation pattern of the material as given by its stressstrain curve. This makes the generation of handbook solutions an expensive task. To a limited extent this task has been realised for a few component congurations, for example, in the Electric Power Research Institute (EPRI) handbook (see below). More generally, however, individual nite element analyses have to be carried out. These analyses require a high level of experience, which is not always available. Therefore, despite the difculties described above, closed form analytical solutions are desired in addition to nite element analyses (Fig. 1). The aim of this paper is to give a brief survey of the basic principles of the more common analytical methods, some of which are discussed in more detail in this special issue. A complete historical review is not attempted. Due to the inherent uncertainties in the determination of the crack tip loading parameter analytical aw assessment methods are aimed at providing conservative results. Consequently, an assessment leading to the result unsafe does not necessarily mean that the component will fail. In those circumstances, a further analysis using a numerical determination of the crack tip loading parameter may be carried out. However, it should be recognised that uncertainties in the transferability of the

0308-0161/00/$ - see front matter q 2001 Elsevier Science Ltd. All rights reserved. PII: S 0308-016 1(01)00008-4

856

U. Zerbst et al. / International Journal of Pressure Vessels and Piping 77 (2000) 855867

Nomenclature a aeff E EH F FY h J Je Jmat Jp Jssy JY K Kmat Kr L Lr Lmax r N n P Po Rm ReL Sr a De d de d mat dY d5 Crack size Plastic zone corrected crack size (different denitions for the EPRI and ETM approaches) Young's modulus E in plane stress; E=1 2 n2 in plane strain Applied load (general for force, moment, pressure, etc.) Limit load Dimensionless function of geometry (EPRI approach) J-integral Linear elastic J-integral Crack resistance in terms of critical J-integral Plastic component of the J-integral Small scale yielding J-integral J-integral at limit load Linear elastic stress intensity factor Crack resistance in terms of critical K Ratio of the applied K to the crack resistance, K/Kmat Characteristic length (EPRI approach) Ratio of applied load, F, to yield load, FY Plastic collapse limit of Lr Strain hardening exponent (ETM denition) Strain hardening exponent (various denitions) Applied load (EPRI approach) Reference load (EPRI approach) Tensile strength Lower yield strength (materials with Luders plateau) Ratio of applied load to plastic collapse load Fit parameter of the RambergOsgood formulation Luders strain CTOD Linear elastic CTOD Crack resistance in terms of critical CTOD CTOD at limit load Specic denition of the CTOD (ETM approach) Strain Yield strain Reference strain (reference stress method) Poisson's ratio Stress Flow strength (usually the average of yield and tensile strength) Reference stress (reference stress method) Yield strength Normalising stress (EPRI approach)

material resistance for fracture in terms of fracture toughness or an R-curve are not resolved in this way. Test specimens, which model the constraint conditions of the component under consideration, are needed to resolve the degree of this uncertainty. 2. Failure assessment diagram or crack driving force? The analytical aw assessment methods pursue two different philosophies, which can be designated as failure assessment diagram (FAD) and crack driving force (CDF) concepts (Fig. 2). In principle, the two concepts can be made fully compatible. However, some of the methods, which will be discussed later, prefer one of both presentations. What is the difference? In the FAD route, a roughly geometry-independent failure line is constructed by normalising the crack tip loading by the material's fracture resistance. The assessment of the component is then based on the relative location of a geometry-dependent assessment point with respect to this failure line. In the simplest application, the component is regarded as safe as long as the assessment point lies within the shaded area below the failure line. It is regarded as potentially unsafe if it is located on the line or outside the shaded area. An increased load or larger crack size will move the assessment point along the loading path towards the failure line. In contrast to the FAD philosophy, in the CDF philosophy the applied and material sides are strictly separated. The determination of the crack tip loading, in the component and its comparison with the fracture resistance of the material are two separate steps. Like the failure line of the FAD concept, the CDF curve can, by suitable normalisation of the load, be made an approximately geometry-independent function, which depends only on the deformation behaviour of the material. 3. Analytical assessment methods 3.1. Introduction Analytical aw assessment methods have been developed for more than 30 years. These are described in numerous published papers but only a limited number can be discussed within this brief survey. In addition, only the basic principles of the various methods will be reviewed. Industrial realisations within guidelines and standards are mentioned only in Table 1 at the end of this paper, but particularly in this eld there have been rapid developments recently, as is apparent from the other papers in this issue. Common to all the methods discussed here is that they are not restricted to a linear elastic deformation pattern of the material but cover the whole range from linear elastic to elasticplastic behaviour (Fig. 3). It is important to note that the limit load of the cracked component plays an important role in almost all the models

e eY e ref n s s  s ref sY so

U. Zerbst et al. / International Journal of Pressure Vessels and Piping 77 (2000) 855867

857

Fig. 1. Principles of analytical determination of the crack tip loading.

discussed below. In fracture mechanics application, the limit load based on the material yield strength marks a transition from contained to net section yielding. However, because of the work-hardening capacity of the material beyond yield this does not correspond to immediate failure. The latter may not happen until the plastic collapse load is reached. Note, in common solid mechanics the distinction between a limit or yield and a collapse load is not usually made because the latter is uniquely dened as the maximum load a component of elasticideally plastic material is able to withstand (Fig. 4).

3.2. Design curve approaches 3.2.1. General principle The so-called design curve approaches are the earliest analytical aw assessment methods. They are based on an idea of Wells in the sixties that was rst realised by Burdekin and Dawes [1] at TWI in 1971. The basic approach is quite simple. It utilises the observation that the crack tip loading in terms of the CTOD (d ) or J-integral depends on the local strain, e , in a quadratic manner in the contained yielding regime and in a linear manner in the net section

Fig. 2. FAD and CDF philosophies of aw assessment.

858

U. Zerbst et al. / International Journal of Pressure Vessels and Piping 77 (2000) 855867

Table 1 Overview on some methods for analytical aw assessment Basic approach Design curves (various approaches mostly for CTOD) Assessment routes FAD CDF Special features and range of application No K factor solution is required where these are based on strain, which is an advantage for shallow crack congurations. Application is usually restricted to shallow cracks (preferably originating from notches) and to limited ligament plasticity. Original FAD approach with the abscissa normalised by the plastic collapse load; should only be used for materials with a yield to ultimate stress ratio close to unity. Fully plastic solutions for the J-integral and the CTOD derived by nite element analyses. The application is restricted to a limited number of component congurations. RambergOsgood description of the stressstrain curve may be a problem. Modied versions for strength mismatch assessment procedures such as the American EWI and the French ARAMIS methods. Based on the reference stress method of Ainsworth, which is a generalisation of the EPRI approach within the frame of FAD philosophy. The R6 document includes numerous appendices for treating weld residual stresses and strength mismatch, constraint effects, probabilistic sensitivity analysis, special applications such as leakbefore-break analysis, etc. Also based on the reference stress method. The procedure includes guidance for leakbefore-break assessment and high temperature creep. The model is part of the comprehensive EFAM methodology of GKSS, which covers toughness testing and failure assessment. In addition to the basic procedure special options for strength mismatch, shallow cracks at notches, and strain-based assessment, guidance for treating data scatter and size effects on toughness are available. Unied European procedure, based on elements of R6 and ETM as well as on additional elements from various sources: BS 7910, the SAQ method and others. SINTAP includes explicitly strength mismatch analysis, statistical treatment of toughness data according to the VTT master curve, compendia for stress intensity factor and limit load solutions as well as for residual stress proles, appendices for constraint effects, reliability analysis and special applications such as leak-beforebreak analysis. Realisation in industrial guidelines British standard BS 7910, assessment level 1, US American and Canadian documents for welded pipelines, API 1104 and CSA Z662, Chinese pressure vessel code CVDA1984; Japanese guideline WES 2805 for aws in fusion welded joints. BSI document PD 6493, now revised to BS7910; `explanation document' of WES 2805. ASME Boiler and Pressure Vessel Code, Section XI, Appendixes G and H for pipelines.

FAD based on Dugdale model

FAD

EPRI-handbook

CDF (FAD)

R6-routine

FAD

Today the most widely applied method, essential part of theBritish standard BS7910, the Draft API 579 procedure of the US American petrochemical industry, the Swedish SAQ method, the French RSE-M Code (2000 Addenda) for nuclear application, and others. A partly modied version is a constituent element of the European SINTAP procedure. Draft document for nuclear industry in France.

CEA-A 16 guide

CDF

ETM

CDF

A modied version is a constituent element of the European SINTAP procedure.

SINTAP

FAD CDF

The procedure is designated for a technical document of the CEN

U. Zerbst et al. / International Journal of Pressure Vessels and Piping 77 (2000) 855867

859

can be written as @ e=eY 2 d 2peY a e=eY 2 0:25

for e=eY # 0:5; for e=eY . 0:5:

In special FAD terms, the failure assessment curve becomes @ e=eY 2 for e=eY # 0:5; dmat 3 FD 2peY a e=eY 2 0:25 for e=eY . 0:5 for ferritic steel applications. Eq. (2) is restricted to ligament plasticity at about the limit load of the cracked component and to shallow cracks. The latter is due to the fact that the crack tip loading is correlated with the applied strain, which is usually dened at the outer surface of the component. The FAD and CDF approaches of Eqs. (3) and (2) are shown in Fig. 5a and b, respectively. Some years later, Dawes proposed a modication of Eq. (2) for deeper cracks [2] and rewrote it in terms of stress in order to t it into the FAD approach of the British PD 6493 [3] procedure. The latter modication is still part of the recently revised version of that document, the British Standard Guide BS 7910 [4]. With some modications, the TWI design curve is also a constituent element of industrial guidelines such as the US and Canadian documents for welded pipelines, API 1104 [5] and CSA Z662 [6], and the Chinese pressure vessel code CVDA-1984 [7]. 3.2.3. The EnJ design curve A J-integral based design curve was proposed by Turner at Imperial College, London, in 1981 [8]. The basic equation of the EnJ method was V ` ef =eY 2 1 1 0:5ef =eY 2 for ef =eY # 1:2; 4 J JY X 2:5 e =e 2 0:2 for e =e . 1:2
f Y f Y

Fig. 3. Application ranges of various fracture mechanics concepts. General approaches cover all ranges within a unied method.

yielding regime. In general terms this may be written as @ C1 e=eY 2 for e=eY , k; 1 d or J eY C2 e=eY 1 C3 for e=eY . k with e Y being the yield strain and C1,2,3 being empirical constants. The quantity k varies within the different design curve approaches from 0.5 to 1. In general terms, C1,2,3 depend on a range of parameters, in particular the geometry and the loading type, for example, bending or tension. For simplicity, these constants were derived for the centre cracked plate loaded under tension, this being expected to provide conservative estimates of the crack tip loading for other congurations. Modern developments in design curves tend to overcome the restriction to tension loading and to incorporate the detailed deformation behaviour of the material. Although solutions for deeper cracks exist, the preferred eld of application is the assessment of shallow cracks, such as those originating from notches. An advantage of the design curve approaches is that they do not require stress intensity factor solutions for the congurations to be assessed. A difculty is that the strain, e , is not clearly dened. Design curves can, in principle, be applied in the frame of the FAD or CDF philosophy (Fig. 5), but are more commonly applied in the CDF framework. 3.2.2. The TWI design curve The most widely applied design curve approach still is that developed by Burdekin and Dawes [1]. In CDF terms it

with e f being the so-called cracked body structural strain. Although Turner gave expressions for e f for some basic congurations, it is not clear how this is dened in general. The quantity JY is the energy release rate, G, calculated elastically, for a load equal to the limit load. 3.2.4. More recent developments Based on extensive nite element analyses, a J-based design curve approach was further developed by a group

Fig. 4. Schematics of limit and plastic collapse loads: (a) in fracture mechanics application, and (b) in common solid mechanics.

860

U. Zerbst et al. / International Journal of Pressure Vessels and Piping 77 (2000) 855867

Fig. 5. Schematics of design curves: (a) special FAD route; (b) CDF route, e.g. according to Eqs. (2) and (3).

at the University of Wales, Swansea. Fitting the material's stressstrain curve by a piece-wise power law @ s=s Y for e # eY; e= eY 5 n s=s Y for e . eY and determining the J-integral from the HRR eld, Lau et al. [9] derived an expression for shallow cracks 0:05 # a=W # 0:1 and for a strain hardening exponent between 2 # n # 30 : V for e=eY # 0:85; b e=eY 2 `  n11=n 6 J JY b C e 2 1:2 X 1JNG for e=eY . 1:2 eY with JNG and C being functions of the relative crack depth, a/W, and the strain hardening exponent, n. J is interpolated by a straight line between 0:85 , e=eY # 1:2: The authors obtained a similar expression for pure bending [10], which has recently been modied for a more generalised stress strain curve t [11]. A largely independent design curve approach was developed by the Japan Welding Engineering Society (WES). Since 1976, the WES 2805 standard and within this frame the Japanese CTOD based design curve has been revised several times. In the most recent version of 1997 [12] it is given by Pp Q eY ae=eY 2 for e=eY # 1 T 2 U U: 7 dT Rp S eY a9e=eY 2 5 for e=eY . 1 8 A special feature of the WES 2805 design curve is that the applied strain, e , is determined from the membrane and bending stress components across the wall and that it depends on the strain hardening of the material if the crack originates from a notch. Recently Xue and Shi from the Beijing Polytechnical University proposed a strain hardening design curve [13] that is based on a simplication of net section plastic CTOD solutions of EPRI (see below). In a CDF format it can be written as ! ! 5n 1 3 s 2 s n11 d 2peY a 10:85 a k 8 8n 1 1 s Y sY with s being the applied stress of the component. The

quantities a and n are the coefcients of the Ramberg Osgood formulation of the stressstrain curve 9 p and k is an out-of-plane constraint factor, being 3=2 for plane strain and 1 for plane stress conditions. Finally, it has to be mentioned that the strain option of the engineering treatment model (ETM) of the GKSS Research Centre in Germany can also be interpreted as a design curve approach (for more details see Ref. [14]). 3.3. Early failure assessment diagrams 3.3.1. General philosophy The FAD for aw assessment was rst used in the socalled R6 approach of the former Central Electricity Generating Board (CEGB) in the UK. An overview on early developments is given in Ref. [15]. The approach recognised that, at one extreme, linear elastic fracture mechanics was applicable and fracture occurred when the stress intensity factor, K, became equal to the fracture toughness, Kmat. At the other extreme failure occurred when the load F reached its value, FL, at plastic collapse. The R6 approach recognised that the use of K beyond the elastic regime underestimated the crack tip loading and, therefore, some plasticity correction was required, as shown in Fig. 6. In general, this correction is a function of the material and

e=e Y s=s Y 1 as=s Y n

Fig. 6. Schematic illustration of the effect of plasticity on crack-tip loading and normalised crack tip loading.

U. Zerbst et al. / International Journal of Pressure Vessels and Piping 77 (2000) 855867

861

the component, the crack geometry and the type of loading. However, in the early R6 approach, it was recognised that by using the two normalising parameters Kr K=Kmat ; and Sr F=FL ; 11 the plasticity correction could be converted to a general purpose failure avoidance curve Kr f Sr : 12 10

In the 1980s, a number of modications of the CEGB FAD were proposed, for example Refs. [17,18]. A disadvantage of the original approach was that it was tailored to elasticideally plastic deformation behaviour due to its origin from the Dugdale model. One aim of the further development was to address in more detail the strain hardening capacity of real materials. An attempt based on a power law description of the stressstrain curve yielded separate FADs for simple bending and tension cases [19]. However, at that time the major interest was focussed on a new development initiated by General Electric in the United States. 3.4. Fully plastic solutions of the EPRI type In 1976, Shih and Hutchinson [20] proposed a new method for determining fully plastic yielding J-integral and CTOD solutions. The method was applied to plate and cylindrical congurations and summarised in the EPRI handbook in 1981 [21]. At the beginning of the nineties it was expanded to a number of additional congurations [22]. Fully plastic solutions were not only generated at EPRI. Alternative examples were given by other authors, for example, solutions for semi-elliptical surface cracks were published by Yagawa and Ueda in 1989 [23]. What is the basis of the method? First, the J-integral (or CTOD) is split into a small scale yielding and a net-section yielding component J Jssy 1 Jp : 16

3.3.2. The early R6 routine of CEGB In the early R6 routine, the problem of a conservative plasticity correction function was solved by choosing the Dugdale model for predicting the elasticplastic crack tip loading. Rewritten in terms of the J-integral this corresponds to !  ! J 8 s 2 p s 2 ln sec 13 Je 2 sY p sY for a defect subject to a remote stress s in an elastic perfectly plastic material, with the linear elastic J-integral, Je, being J e K 2 =E: 14 In a second step, the elasticplastic J in Eq. (13) was replaced at failure by the fracture resistance of the material, Jmat, which is related to Kmat by an equation similar to Eq. (14). The applied load normalised by the plastic collapse load, dened for a ow stress s ; then replaced (s /s Y) in Eq.  (13). This led to a geometry independent failure line, which formed the rst FAD of CEGB [16] (Fig. 7). In terms of Eq. (12), the Dugdale plasticity correction corresponds to  !21=2 8 p Sr ln sec : 15 f S r Sr 2 p2 The FAD of Eq. (15) could then be interpreted simply as an interpolation between the two limiting failure states brittle fracture and plastic collapse of a cracked component.

The plastic part, Jp, is derived from dimensional arguments recognising that for proportional loading, the fully plastic component must be related to the applied load, which in the EPRI handbook is usually designated as P, by  n11 P : 17 Jp aeo s o hL Po The load is normalised by a reference load, Po, which, in principle, may be freely chosen provided it is proportional to s o but is often identied with the limit load of the cracked component. The quantities e o, s o, a and n are the tting parameters of the RambergOsgood formulation of the stressstrain curve (Eq. (9) is used with e Y and s Y replaced by e o, s o). Like Po the characteristic dimension L can be freely chosen. Usually it is identied with the ligament length or the crack size. The derivation of the non-dimensional parameter h is discussed below. The small scale yielding component of J, Jssy, is derived from a plastic zone correction to K J K 2 aeff =E H 18

Fig. 7. FAD according to R6-Rev. 1, schematically.

for an effective crack length that depends on strain hardening   1 1 n21 K 2 : 19 aeff a 1 1 1 P=Po 2 bp n 1 1 s o

862

U. Zerbst et al. / International Journal of Pressure Vessels and Piping 77 (2000) 855867

which extensive tables of h factors are available. A critical point is the RambergOsgood t to the stressstrain curve. While this is convenient for nite element calculations, most materials do not follow this t satisfactorily and in particular the important region near the yield strength is usually poorly described. 3.5. Approaches based on the CEGB reference stress method 3.5.1. The reference stress method Some limitations of the EPRI handbook were overcome by the development of the reference stress method which was rst applied to elasticplastic fracture by Ainsworth in 1984 [25]. In fact, the method has been used earlier by Ainsworth [26,27] to apply steady state creep fracture mechanics in the absence of detailed numerical solutions such as those in the EPRI handbook. With the production of the handbook, it was possible to examine the reference stress method in more detail to develop a general FAD approach. Two modications were essential: First the Jp solution according to Eq. (17) was replaced by the expression   s e 20 J p s ref hL eref 2 ref o : so Dening the reference stress, e ref, by   P s ref s o Po

Fig. 8. Basic principle of the EPRI approach.

The quantity b is 6 for plane strain and 2 for plane stress conditions. The choice of s o is arbitrary for the purposes of the RambergOsgood t to the stressstrain curve and the subsequent use in Eq. (17) but to be of relevance to Eq. (19) it must be of the order of the yield strength. J-integral values were determined for different sets of stressstrain curves as dened by Eq. (9) using extensive 2D nite element analyses. The component and crack dimensions were varied step-wise and the fully plastic results were tted by Eq. (17) to dene the quantity h which of course depended on the choice of Po and L (Fig. 8). Tables for this parameter as a function of component, crack size and strain hardening generated for a range of plate and cylindrical congurations form the main items of the EPRI handbooks. In addition to J solutions, solutions for CTOD and displacement are available. In 1980, Bloom proposed that the EPRI solutions were replotted in a FAD format [24] (Fig. 9). There were, however, two differences to the CEGB FAD. The applied load was normalised by the EPRI normalising load (cf. Section 3.1) proportional to s o and not by the collapse load proportional to s : The failure curve became a function  of the strain hardening of the material. Both features were direct consequences of the EPRI approach. The main advantage of the EPRI handbook is that it is based on nite element analyses. There are, however, problems in its application. One limitation of the handbook is that it is applicable to only a few congurations, those for

21

and a reference strain, e ref, such that it corresponds to s ref via the RambergOsgood formulation. Eq. (20) is identical with Eq. (17). It was, however, possible to make the method much more general. Introducing s ref and e ref point-wise from the real stressstrain curve into the analysis allowed an exact description of any material. However, the quantity h in Eq. (20) was still a function of the strain hardening exponent, n. This dependency was minimised by redening the reference load, Po, in Eq. (21). Having done this h could simply be set equal to the value for n 1; the corresponding value for linear elastic material behaviour, which could be related to K. In this way Ainsworth reached the expression   mK 2 Eeref 21 22 Jp E s ref with m equal to 0.75 for plane strain and 1 for plane stress, respectively. Eq. (22) allowed the determination of the plastic component of the J-integral simply from the linear elastic K-factor, which implicitly took into account the component geometry. The deformation behaviour was reected by the stressstrain curve from which the reference stress and strain were taken from. To determine the reference stress as described above, it was necessary to modify the value of EPRI reference load Po and this required the numerical h solutions. However, it was found that the modied reference load was often close to the yield load and, therefore, the

Fig. 9. EPRI based FAD according to Bloom [24].

U. Zerbst et al. / International Journal of Pressure Vessels and Piping 77 (2000) 855867

863

reference stress was redened as

s ref Lr s Y

F s : FY Y

23

The quantity Lr, based on yield stress, rather than the owstress based quantity Sr of Eq. (11) forms the basis of the R6 revision 3 method, which is described in the following sections. Minor additional corrections were introduced at that stage to ensure conservatism. The overall accuracy of the reference stress method and indeed the accuracy of the EPRI method used in its development were also examined [28]. It was concluded that the limit load based reference stress was acceptably accurate for practical applications and generally erred on the side of conservatism. 3.5.2. The FAD according to R6 Rev. 3 The R6 routine, rev. 3 [29], is probably the most widely applied analytical aw assessment method worldwide. It forms sometimes slightly modied the basis of industrial codes and guidelines such as the British standard BS7910 for assessing aws in weldments [4], the draft API 579 tness for service procedure of the American petrochemical industry [30], the Swedish SAQ method [31] and the French RSE-M Code (2000 Addenda) for assessing aws in nuclear power components [32], and others. Introducing some additional reductions, the reference stress method according to Eq. (22) was rewritten in terms of the R6 parameters of Eqs. (10)(12) but using Lr of Eq. (23) to replace Sr to give a failure avoidance curve [25] !2 1 2 1 : 24 f Lr Eeref =s ref 1 L2 =1 1 L2 r r 2 The second term in Eq. (24) can be interpreted as a plasticity correction in the intermediate small scale yielding region where the bulk behaviour of the component is still elastic but the ligament ahead of the crack already experiences substantial local plasticity. In order to improve the t to real conditions, the denominator 1 1 L2 was replaced r semi-empirically by the term (Ee ref/s ref) [29] to give the equation 4 5 Eeref L3 s Y r 1 ; 25 f L r Lr s Y 2Eeref which is termed the material specic option 2 of R6 rev.3. The application of Eq. (25) requires knowledge of the complete stressstrain curve. In particular the region around the yield point has to be available in a detailed manner. However, there are many cases where this information is not available to the user. Therefore, Eq. (25) was applied to a number of materials to generate a material independent lower bound curve, which is the more conservative option 1 curve of R6 rev. 3 f Lr 1 2 0:14L2 0:3 1 0:7 exp20:65 L6 : r r 26

Fig. 10. FAD according to R6-Rev. 3, schematically.

cally. Note, that in contrast to the original CEGB FAD the ligament plasticity is referred to the limit load and not to the collapse load of the cracked component. As a consequence, a limit line against plastic collapse had to be introduced separately. In the gure this is marked as Lmax : Usually r this is determined as Lmax r

s  1 with s s Y 1 Rm  sY 2

27

although different denitions for special classes of materials also exist in R6. Currently a fourth revision of the R6 routine is in preparation, which includes an update with respect to the recently developed European aw assessment method SINTAP (see below). 3.5.3. The Js method of CEA Another method based on the reference stress approach is the so-called Js method, which was rst proposed by Moulin et al. [33] in 1993. It became a constituent part of the French CEA-A16 guide [34] in 1997. With the elastic part of J estimated as Je mK 2 =E Eq. (22) becomes   Eeref J Je 1 Jp Je s ref 28

29

for plane stress conditions m 1: J is the complete J-integral, in A16 designated as Js. 3.6. The engineering treatment model The ETM has been developed at the GKSS Research Centre in Germany since the mid eighties [35]. In 1997/ 1998, the method was provided both for homogeneous components and for heterogeneous congurations such as strength-mismatched welded joints in handbook format [36,37]. As with the development of the EPRI and reference stress approaches, the motivation was to derive a fracture mechanics based aw assessment method including the effects of material hardening on crack tip loading.

In Fig. 10 both R6 failure curves are plotted schemati-

864

U. Zerbst et al. / International Journal of Pressure Vessels and Piping 77 (2000) 855867

Fig. 12. ETM curve, J-integral option, schematically.

Fig. 11. Basic principle of the ETM.

The basic concept realised in the ETM is shown in Fig. 11. The deformation behaviour of the ligament ahead of the crack is assumed to be satisfactorily described by the uniaxial tensile test tted by the piece-wise power law @ for s , s Y Ee s : 30 N s Y e=eY for s $ s Y In Eq. (30) the strain hardening exponent is designated by a capital N in order to distinguish it from the n ( 1/N) used in Eq. (5). More importantly, it is dened from the engineering stressstrain curve, whereas all the methods reviewed above have used the true stressstrain curve for describing the deformation behaviour of the material. Another feature of the ETM is that it prefers a special denition for the CTOD the CTOD-d5 which is locally measured at the crack tip (see Ref. [14]). The J-integral is applicable within the ETM approach too. The ETM curve which, like the R6 FAD curve, is an approximately geometry independent master curve is given by   b 1 F 2 d5 1 K 1 Keff 31 for F , FY E mEs Y FY and F FY !1=N

tip loading at the limit load, FY, the coefcients b 1 and m have constant values b1 2:09 and m 2 for plane strain and b1 2:41 and m 1 for plane stress conditions. Keff is the plastic zone corrected stress intensity factor. The method follows exclusively the CDF philosophy, i.e. the ETM curve represents the crack tip loading. Failure conditions are given by the intersection of the ETM curve with the fracture resistance of the material, represented by Jmat or d 5Ymat, see Fig. 12. In addition to the basic ETM, special options have been derived for shallow cracks originating from notches and for strain-based assessment, which are discussed in more detail in Ref. [14]. An update with respect to the European aw assessment method SINTAP (see below) is planned. 3.7. The European aw assessment method SINTAP The SINTAP procedure [38] is the outcome of a recently nished BriteEuram project in which seventeen institutions from nine European countries were participants. The procedure offers both a FAD and a CDF route. They are complementary and yield identical results. The basic failure line of the FAD route is Kr f Lr 35 with Kr being K/Kmat as in Eq. (10) for primary loading and Lr being F/FY as in Eq. (23). The CDF expressions for primary loading are J Je f Lr 22 36

d5 d5Y

for F $ FY ;

32

with Je K 2 =E H for the J-integral and 37

or in terms of the J-integral by 1 2 J Keff E and J JY F FY for F , FY !11N=N 33

d de f Lr 22
with

38

for F $ FY ;

34

de K 2 =E H s Y

39

respectively. The quantities d 5Y and JY designate the crack

for CTOD. The effective Young's modulus, E H , is E/ (1 2 n 2) for plane strain and E H E for plane stress

U. Zerbst et al. / International Journal of Pressure Vessels and Piping 77 (2000) 855867

865

Fig. 13. Overall structure of the SINTAP procedure.

conditions, respectively. It is of importance that the function f(Lr) is identical for the FAD and CDF routes. In Fig. 13, the overall structure of the procedure is presented. SINTAP consists of various analysis levels which are constituted by the quality and completeness of the input information required. Higher levels are more complex than the lower ones and need improved input information; however, the user is `rewarded' by less conservative results. An unacceptable result does not necessarily mean failure of the component analysed. Instead, it rather provides a motivation for repeating the analysis at the next higher level. On the other hand, if a lower level analysis indicates safety the user does not need to provide improved input information and to perform a more complex analysis. The SINTAP standard analysis levels are: Level 1. This forms the basic level. Only the toughness and the yield strength and the ultimate tensile strength of the material need to be known. The function f (Lr) includes elements from R6, rev. 3 and the ETM, both being slightly modied in order to provide compatible FAD and CDF solutions. A SINTAP specic feature is that it offers different sets of equations for materials with and without Luders plateau. This is an important modication of both original methods. Level 2. The second level is a modication of level 1 for inhomogeneous congurations such as strengthmismatched weldments which is also based on R6, rev. 3 and the ETM. Combinations of materials where both, none or only one showing a Luders plateau can be analysed.

Level 3. This level is identical to R6 option 2 which requires toughness data and the complete stressstrain curve of the material. Both homogeneous and strength mismatch components can be handled, the treatment of strength mismatch being based on the ETM mismatch handbook [36] and on Appendix 16 of R6 [39]. There are additional levels: Level 0. This is a default option which requires the yield strength of the material only and which allows the fracture resistance of the material to be estimated from Charpy data. A level 0 analysis will usually be highly conservative but it enables a fracture mechanics analysis even with a minimum of input information a case that is not unusual in industrial practice. The advanced levels are: Level 4. Within this level, loss of constraint in thin sections or predominately tensile loading is considered. Level 5. This is based on complete numerical analysis, e.g. nite element analysis. Level 6. This is a special option for evaluating leak before break in pressurised components. The SINTAP procedure is described in more detail elsewhere in this issue [40]. Here, only the level 1 equations for materials showing a Luders plateau will be given for

866

U. Zerbst et al. / International Journal of Pressure Vessels and Piping 77 (2000) 855867

illustration. These are !21=2 1 ; f Lr 1 1 L2 2 r 1 f 1 l 1 2l with EDe ReL !21=2

References for 0 # Lr , 1; for Lr 1 40


[1] Burdekin FM, Dawes MG. Practical use of linear elastic and yielding fracture mechanics with particular reference to pressure vessels. Inst Mech Eng Conf, Lond 1971:2837. [2] Dawes MG. The CTOD design curve approach: limitations, nite size and application. The Welding Institute, TWI-Report 780302/85/4613, 1985. [3] PD 6493: Guidance on methods for assessing the acceptability of aws in fusion welded structures. London: BSI, 1991. [4] BS 7910: Guide on methods for assessing the acceptability of aws in fusion welded structures. London: BSI, 1999. [5] American Petroleum Institute. API 1104: welding of pipelines and related facilities. Appendix A: alternative acceptance standards for girth welds. American Petroleum Institute, 1994. [6] Canadian Standards Association. CSA Z662-94: oil and gas pipeline system. Appendix K: standards for acceptability of circumferential pipe butt welds based on fracture mechanics principles. Canadian Standards Association, 1994. [7] Chinese Society of Pressure Vessel Technology and Society of Chemical Machinery Engineering. CVDA-1984: code of assessment of pressure vessels, 1984. [8] Turner CE. A J-based engineering usage of fracture mechanics, vol. 3. ICF 5, Cannes 1981;122. [9] Lau CE, Lee MMK, Luxmoore AR. Methodologies for predicting J-integrals under large plastic deformation I. Further developments for tension loading. Engng Fract Mech 1994;49: 33754. [10] Lau CE, Lee MMK, Luxmoore AR. Methodologies for predicting J-integrals under large plastic deformation II. Single edge notch specimens in pure bending. Engng Fract Mech 1994;49:35569. [11] Boothman DP, Lee MMK, Luxmoore AR. A shallow crack assessment scheme for generalised material behaviour in bending. Engng Fract Mech 1997;57:493506. [12] Japan Welding Engineering Society. WES 2805-1997: Method of Assessment for Flaws in Fusion Welded Joints with Respect to Brittle Fracture and Fatigue Growth, 1997. [13] Xue H, Shi Y. CTOD design curve in consideration of material strain hardening. Int J Pres Ves Piping 1998;75:56773. [14] Schwalbe K-H, Zerbst U. The Engineering Treatment Model. Int J Pres Ves Piping 2001;77(1415):90316. [15] Larsson LH. Use of EPFM in design. In: Larson LH, editor. Advances in elasticplastic fracture mechanics, 1979. London: Applied Science Publishers, 1979. p. 261300. [16] Harrison RP, Milne I, Loosemore K. Assessment of the Integrity of Structures Containing Defects. CEGB Report R/H/R6-RRev1, 1977. [17] Rich TP, Ghassem MM, Cartwright DJ. Fracture diagrams for cracked stiffened panels. Engng Fract Mech 1985;21:100517. [18] Will P, Michel B, Zerbst U. JTJ-controlled crack growth modication of JR testing and failure assessment diagrams. Engng Fract Mech 1987;28:197201. [19] Kaiser S. A failure assessment diagram for strain hardening materials under cyclic and monotonic loading. Engng Fract Mech 1984;20: 5163. [20] Shih CF, Hutchison JW. Fully plastic solutions and large-scale yielding estimates for plane stress crack problems. ASME J Engng Mat Technol 1976;98:28995. [21] Kumar V, German MD, Shih CF. An engineering approach for elasticplastic fracture analysis. EPRI-Report NP-1931, EPRI, Palo Alto, 1981. [22] Zahoor A. Ductile fracture handbook. Novotech Cop and EPRI, Res Proj 1757-69, vol. 1: 1989; vol. 2: 1990; vol. 3: 1991. [23] Yagawa G, Ueda H. Behaviour of surface crack in plates subjected to tensile loads: analysis based on fully plastic solutions. Nucl Engng Design 1989;111:18996.

41

l11
and

42

N21=2N f Lr f Lr 1Lr

for 1 # Lr , Lmax : r

43

Eq. (41) leads to a step change in f(Lr) in the Luders plateau. If the Luders strain is unknown it can be estimated conservatively from an empirical correlation ! R 44 De 0; 0375 1 2 eL : 1000 The strain-hardening exponent can also be determined on a conservative, empirical basis: ! ReL 45 N 0; 3 1 2 Rm with ReL being the lower yield strength and Rm the tensile strength of the material. The plastic collapse limit is dened as ! 1 ReL 1 Rm Lrmax : 46 2 ReL The SINTAP procedure has been developed as a contribution towards the development of a European Committee for Standardisation (CEN) tness for service standard. Presently it is planned as an input to a technical document of the CEN. 4. Conclusions The basic principles of various analytical aw assessment methods such as the design curve approaches of TWI and WES, the EPRI handbook, the R6 routine, the CEA-A16 guide, the ETM method of GKSS and the recently developed European SINTAP procedure have been reviewed. The aim has been to give an impression of the historical development of all these methods and to point to some common features, which may be important in the context of any future development towards an international unied aw assessment method. Acknowledgements This paper is published with permission of GKSS Research Centre and British Energy Generation Ltd.

U. Zerbst et al. / International Journal of Pressure Vessels and Piping 77 (2000) 855867 [24] Bloom JM. Prediction of ductile tearing using a proposed strain hardening failure assessment diagram. Int J Fract 1980;6:R737. [25] Ainsworth RA. The assessment of defects in structures of strain hardening materials. Engng Fract Mech 1984;19:63342. [26] Milne I, Ainsworth RA, Dowling AR, Stewart AT. Assessment of the integrity of structures containing defects. CEGB Report R/H/R6Revision 3. Latest ed. 1986; latest ed. British Energy, 1999. [27] Ainsworth RA. Some observations on creep crack growth. Int J Fract 1982;20:14759. [28] Ainsworth RA. The initiation of creep crack growth. Int J Solids Struct 1982;18:87381. [29] Miller AG, Ainsworth RA. Consistency of numerical results for power-law hardening materials and the accuracy of the reference stress approximation for J. Engng Fract Mech 1989;32:23347. [30] American Petroleum Institute. API 579. Recommended practice for tness-for-service, Draft, 1996. [31] Bergman M, Brickstad B, Dahlberg L, Nilsson F, Sattari-Far I. A procedure for safety assessment of components with cracks handbook. SA/FoU Report 91/01, AB Svensk Anlaggningsprovning, Swedish Plant Inspection Ltd., 1991. [32] RSE-M Code, 1997 ed. and 2000 Addenda: Rules for Inservice Inspection of Nuclear Power Plant Components. AFCEN, Paris.

867

[33] Moulin D, Nedelec M, Clement G. Simplied method to estimate J development and application. SMIRT 1993;12(G04/4):1516. [34] Commissariat a l'Energie Atomique (CEA). A16: Guide for defect assessment and leak before break analysis, Draft, 1999. [35] Schwalbe K-H, Cornec A. The engineering treatment model (ETM) and its practical application. Fatigue Fract Engng Mat Struct 1991;14:40512. [36] Schwalbe K-H, Kim Y-J, Hao S, Cornec A, Kocak M. EFAM, ETM MM 96: the ETM method for assessing the signicance of crack-like defects in joints with mechanical heterogeneity (strength mismatch). GKSS Report 97/E/9, 1997. [37] Schwalbe K-H, Zerbst U, Kim Y-J, Brocks W, Cornec A, Heerens J, Amstutz H. EFAM ETM 97: the ETM method for assessing crack-like defects in engineering structures. GKSS Report GKSS 98/E/6, 1998. [38] SINTAP: structural integrity assessment procedure for european industry. Final Procedure, 1999. BriteEuram Project No. BE951426, British Steel. [39] R6-Revision 3, Appendix 16. Allowance for strength mismatch effects, British Energy, 1997. [40] Ainsworth RA, Bannister AC, Zerbst U. An overview of the European aw assessment procedure SINTAP and its validation. Int J Pres Ves Piping 2001;77(1415):86976.

S-ar putea să vă placă și