Mohamad J Cheaitani
Paper presented at 26th International Conference on Offshore Mechanics and Arctic Engineering, OMAE 2007, San Diego, California, 10-15 June 2007. Paper no. 29558.
The use of an engineering critical assessment (ECA) approach to derive flaw acceptance criteria for pipe girth welds has become common practice. It allows the maximum tolerable size of weld flaws to be determined on a fitness-for-purpose basis, offering substantial advantages over the conventional workmanship approach.
BS 7910:2005 is widely used to derive ECA-based flaw acceptance criteria for pipe girth welds. It offers a flexible assessment framework within the context of the well-established failure assessment diagram (FAD) approach. However,it can be relatively complex to apply and it may lead to assessments that are more conservative than codified pipeline-specific procedures.
This paper illustrates, through practical case studies on assessing the significance of circumferential girth weld flaws, some of the options available to the user of BS 7910. The case studies cover the selection of the FAD(generalised or material-specific, with and without yield discontinuity), tensile properties (specified minimum or actual values); fracture toughness properties (single point CTOD values including δ 0.2BL and δ m , or full CTOD resistance R-curve), and welding residual stress (assumed to be uniform through the pipe wall with a yield strength magnitude, or considered to have a through-wall distribution associated with aspecific level of welding heat input).
The engineering critical assessment (ECA) approach is widely used to derive flaw acceptance criteria for pipeline girth welds. This allows the maximum tolerable size of weld flaws, including surface breaking and embedded circumferential planar flaws, to be determined on a fitness-for-purpose basis using recognised and well-tried fracture mechanics methods. A typical ECA involves assessing the significance of such flaws with regard to failure mechanisms that the pipeline may experience during construction, commissioning and service. These failure mechanisms include fracture and plastic collapse, under static loading conditions, and fatigue under cyclic loading conditions.
The most commonly used approach to assess the significance of flaws with regard to fracture and plastic collapse is the failure assessment diagram (FAD), which involves the calculation of a fracture parameter, equal to the ratio of the elastic crack driving force to the material's fracture toughness, and a plastic collapse parameter defined as either the ratio of applied load to the limit load or, equivalently, the ratio of the reference stress to the yield strength. The reference stress characterises the increase in stress in the vicinity of a flaw due to the presence of the flaw. The fracture and plastic collapse parameters are represented on a vertical axis and a horizontal axis,respectively. The axes are joined by a failure assessment curve, which incorporates the effect of plasticity on crack driving force, Figure 1. If the assessment point, corresponding to the fracture and plastic collapse parameters, falls within the area bounded by the axes and the failure assessment curve, the flaw is considered acceptable, otherwise the flaw is deemed unacceptable, ie it could lead to failure.
Fig.1. A typical failure assessment diagram
The FAD approach was initially developed by the former UK Central Electricity Generating Board (CEGB) and formed the backbone of its R6 assessment procedure ( BEGL, 2006). Since 1991, the FAD approach has become the main fracture assessment tool in PD 6493:1991 ( BSI, 1991) and its successor, BS 7910:1999 ( BSI, 2000), largely replacing the CTOD design curve approach, first published in PD 6493:1980 ( BSI, 1980). The latest Edition of BS 7910 ( BSI, 2005) is, similarly to its predecessors, widely used to assess the significance of flaws in welded metallic structures, including the derivation of ECA-based flaw acceptance criteria for circumferential girth weld flaws. It provides guidance on fracture assessments using the FAD approach and fatigue assessments using the Paris Law. The guidance is thorough and versatile, but can be relatively complex to apply since it requires a high level of proficiency in fracture mechanics and a multi-disciplinary understanding of the parameters required to conduct an assessment. In comparison with pipeline-specific ECA procedures, such as those of API 1104 ( API, 1999) and EPRG ( Knauf and Hopkins, 1996), BS 7910 offers more versatility and allows a wider range of conditions (including joint and flaw geometry, loading type, material properties, and failure mechanisms) to be assessed. However, as a result of its versatility, BS 7910 is more prone to subjective interpretations and, consequently, different users could obtain significantly different answers to the same problem.
This paper illustrates, through practical case studies on assessing the significance of circumferential girth weld flaws and/or derivation of flaw acceptance criteria for such flaws, some of the options available to the user of BS 7910. The case studies cover the selection of FAD (generalised or material-specific, with and without yield discontinuity), tensile properties (specified minimum or actual values); fracture toughness properties (single point CTOD including δ 0.2BL and δ m , or full CTOD resistance R-curve), and welding residual stress (assumed to be uniform through the pipe wall with a yield strength magnitude, or considered to have a through-wall distribution associated with a specific level of welding heat input).
The above parameters and assessment models are evaluated using case studies, which are built around the following problem (referred to hereafter as the Base Case). The case studies are largely similar to the Base Case except with regard to some input parameters, which are modified in order to illustrate relevant analysis options. Where appropriate, alternative more advanced models are highlighted.
The base case is aimed at estimating the minimum fracture toughness (expressed in terms of CTOD) required to tolerate a 3x50mm circumferential girth weld flaw under a range of axial installation stresses. The minimum CTOD is defined as that producing an assessment point on the failure assessment line, which separates the FAD safe and unsafe regions. The flaw is assumed to be surface breaking, have a semi-elliptical shape and be located on the fusion boundary of the girth weld on the pipe outer surface. The pipe is assumed to have an outside diameter of 273mm and a wall thickness of 19.1mm. Its yield and tensile strengths are assumed to be equal to the specified minimum yield strength (SMYS) and tensile strength (SMTS) of API 5L Grade X65 steel (448 and 530MPa, respectively). The weld tensile properties are assumed to overmatch those of the pipe. The pipe is subjected to a maximum axial stress ranging from 0.75xSMYS to 1.05xSMYS, which is assumed to be uniform through the pipe wall. The stress intensity factor is calculated using the Raju and Newman equations recommended in BS 7910 for surface flaws in flat plates. The stress intensity magnification factor, which reflects the influence of the weld geometry on stress distribution in the vicinity of weld toe flaws, is quantified using a 2D MK solution assuming that the weld cap width is equal to the pipe wall thickness. Welding residual stresses, initially of yield strength magnitude, and transverse to the weld, ie parallel to the pipe axis, are assumed to be present in the region containing the flaw. The magnitude of such stresses is controlled by the lower strength material, which in this case is the pipe metal. Consequently, a welding residual stress equal to the pipe SMYS, assumed to be constant through the wall thickness, is used in the assessments. This initial value is allowed to relax as the applied stress is increased according to Clause 126.96.36.199 in BS 7910.
It is worth noting that the above approach would also apply to assessing the significance of a surface breaking flaw located in the weld metal, except that no Mk factor would be needed since the flaw is not located at the weld toe.
Choice of FAD
There are several options for defining the failure assessment curve, which joins the fracture axis to the plastic collapse axis within a FAD, depending on the available data and desired accuracy. Greater accuracy may be achieved if the failure assessment curve is based on the actual stress-strain curve of the material, or if it is constructed using J-integral data from elastic-plastic finite element analyses. The effects of the choice of FAD are illustrated using the Base Case and the following FADs, which are shown in Figure 2:
- Generalised Level 2A FAD for continuous yielding (BS 7910 Clause 188.8.131.52)
- Generalised Level 2A FAD for discontinuous yielding (BS 7910 Clause 184.108.40.206)
- Material-specific Level 2B FAD for continuous yielding (BS 7910 Clause 220.127.116.11)
- Material-specific Level 2B FAD for discontinuous yielding (BS 7910 Clause 18.104.22.168)
Fig.2. Generalised and material-specific FADs considered
The generalised Level 2A FADs require a knowledge of the yield and tensile strength of the pipe material (assumed here to be equal to 448 and 530MPa respectively) and whether the stress-strain response of the pipe includes a yield discontinuity (usually found in seamless pipe but not in larger diameter seam-welded pipe). The material-specific Level 2B FADs have been constructed using stress strain curves, with and without a yield discontinuity, Figure 3. The curves have been constructed such that they have a yield strength of 448MPa (defined as the stress at 0.5% total strain) and a tensile strength of 530MPa (reached at about 7.5% total strain). The curve with discontinuous yielding has been smoothed around the intersection between the linear elastic part and Lüders' plateau, which extends up to around 1.8% strain. The two curves are identical for stresses less than 350MPa and for strains greater than 5.8%. Although theoretical, the curves have realistic shapes, which have been observed in tensile tests on offshore pipeline steels.
Fig.3. Stress-strain curves used to construct the material-specific FADs
Figure 2 shows that for 0.65 < Lr < 1.0, the smallest allowable √ δr (or K r ) values are those of the Generalised Level 2A FADs followed by the material-specific FAD for continuous yielding, then by the material-specific FAD for discontinuous yielding. For L r > 1.0, the smallest allowable √ δr (or K r ) values are those of the Generalised Level 2A FAD for discontinuous yielding, followed by the material-specific FAD for discontinuous yielding, then the generalised FAD for continuous yielding and material-specific FAD for continuous yielding. An almost mirror-reflection of these observations can be seen in the results in Figure 4, which are expressed in terms of minimum required CTOD vs. the ratio of maximum installation stress (P m ) to SMYS. This is not surprising since the shape of the failure assessment curve reflects the shape of the effective elastic-plastic CTOD crack driving force. Consequently, for a given Pm/SMYS value, an analysis based on the FAD with the smallest allowable √ δr (or K r ) will give the largest required CTOD and vice versa. Figure 4 shows that for P m /SMYS < 0.97 (which corresponds to L r < 1.0), the largest required CTOD is obtained from the analysis based on the Generalised Level 2A FADs followed by the material-specific FAD for continuous yielding, then by the material-specific FAD for discontinuous yielding. For P m /SMYS > 0.97, the largest required CTOD is obtained from the analysis based on the Generalised Level 2A FAD for discontinuous yielding, followed by the material-specific FAD for discontinuous yielding, thenthe generalised FAD for continuous yielding and material-specific FAD for continuous yielding.
Fig.4. Results illustrating the effects of selection of FAD on minimum required CTOD
The above results show that the material-specific Level 2B FAD provides less conservative and significantly more favourable results than the generalised FADs. However, care should be exercised when constructing a material-specific FAD for the assessment of pipeline girth welds. Several stress-strain curves, obtained from longitudinal specimens with data recorded at least up to the tensile strength, are likely to be needed to ensure that the variations in pipe properties around the circumference and between different heats are taken into account. The curve used to construct the FAD should be defined such that it provides a lower bound to all other curves. Typically, for pipeline girth welds subjected to applied stresses up to SMYS, the curve is based on the mean minus two standard deviations of tensile properties taken from a large number of pipe mill test certificates. The FAD cut-off, L r max , may be defined based on the mean of the yield and tensile strengths of the pipe material (which is permitted according to Clause 7.3.1 of BS 7910).
Further research is needed to assess the effects of yield discontinuity on effective elastic-plastic crack driving force (and corresponding FAD) when the pipe is loaded by bending. In this case, the yield discontinuity, which occurs in tension-loaded specimens, may have less effects than those implied by the material-specific FAD.
For a given applied stress, the plastic collapse parameter, L r
, is inversely proportional to the yield strength. An increase in the yield strength is therefore beneficial since it leads to a decrease in L r
and consequently a larger allowable √ δr (or K r
). When assessing pipe girth welds, more favourable results may be obtained if the actual yield strength of the pipe rather than its SMYS is used since the former should be greater than the latter. The effects of using actual tensile properties are illustrated by using the Base Case with two additional analyses where the actual yield and tensile strengths of the pipe are assumed to be 5 and 10% higher than the specified minimum values for API Grade X65 steel. In each case, it is assumed that the strength of the weld metal overmatches that of the pipe metal. Results from these analyses, conducted using the Generalised Level 2A FAD for continuous yielding and expressed in terms of minimum required CTOD, are shown in Figure 5
. This indicates that approximately 14 and 24% reductions in required CTOD are achieved when the tensile properties of the pipe are increased by 5 and 10% above the specified minimum values, respectively.
Fig.5. Results illustrating the effects of using specified minimum and actual tensile properties (5 and 10% above specified minimum values) on minimum required CTOD
The above case study shows that more favourable results are generally obtained when using actual tensile properties rather than specified minimum values. When deriving flaw acceptance criteria for pipeline girth welds subjected to applied stresses up to SMYS, it is recommended to use the lower bound yield strength (defined as the mean minus two standard deviations of normally distributed yield strengths) to specify the parameter L r .
The above approach can be refined further if overmatching is taken into account, since in this case it may be possible to justify the use of a lower L r value than that obtained using the pipe yield strength. However, there are no published generalised mismatch limit load solutions (that are required to define a mismatch value of L r ) for circumferential finite length part-thickness flaws in cylinders.
Level 2 assessments use a single-point value of the fracture toughness, which is taken as the lowest of δ c , δ u or δ m when expressed in terms of CTOD or the lowest of J c , J u or J m when expressed in terms of J. These data are determined from standardised single edge notch bend (SENB) specimens tested in accordance with BS 7448 Part 2 ( BSI, 1997a) or equivalent standards. δ c and J c results (respectively δ u and J u results) are obtained when unstable fracture occurs after less than (respectively greater than) 0.2mm of crack extension prior to reaching a maximum load on the load deflection curve. δ m and J m results are obtained when a maximum load is reached on the load deflection curve of the specimen, usually after a certain amount of crack extension by ductile tearing. This tearing, denoted here as Δa m , can be determined if the full resistance curve (R-curve) is known, eg from tests to BS 7448 Part 4 ( BSI, 1997b) or equivalent standards. BS 7910 permits the use of a maximum load value, δ m or J m , in a Level 2 assessment for components where stable tearing is acceptable. In this case, the assessment is based on the initial flaw dimensions (ie the flaw size before application of load), although the toughness used represents the material resistance after Δa m of ductile crack extension. Such an approach introduces an error into the assessment since it amounts to underestimating the crack driving force or equivalently overestimating the fracture toughness (the effective crack driving force for the flaw increases as its size increases by Δa m ), which could potentially produce unsafe results. The error resulting from such an assessment is dependent on the ratio Δa m /a, where 'a' represents the height (through-thickness dimension) of the flaw under consideration, and the ratio δ m / δ 0.2BL where δ 0.2BL is an engineering definition of the toughness at initiation of crack extension (defined as the CTOD at 0.2mm crack extension offset to the blunting line). If Δa m /a is small (eg < 0.1) and δm/ δ 0.2BL is also small (eg < 1.2), the error will be small and vice versa. However, in practice the potential non-conservatism resulting from this error is likely to be compensated for by several sources of conservatism within the assessment procedure. The most obvious compensating factor is that toughness data obtained from standardised deeply cracked SENB specimens are generally lower bound estimates of the true fracture toughness of the material (see last paragraph in this section). Another compensating factor is that the fracture toughness of the material (after tearing to levels > align="left" Δa m ) will be higher than δ m (or J m ). Where no significant tearing is permitted, BS 7910 recommends that the fracture toughness be taken as δ 0.2BL or J 0.2BL respectively, which are derived from R-curve tests to BS 7448 Part 4 (J 0.2BL is J at 0.2mm crack extension offset to the blunting line). For a tearing analysis at Level 3, a full tearing resistance curve derived from either δ or J measurements is needed.
The effects of the choice of fracture toughness (single-point value or a full R-curve) is illustrated in a case study, which is similar to the Base Case except that it is aimed at estimating the maximum installation stress that can be tolerated assuming a range of fracture toughness values expressed in terms of single-point CTOD ( δ 0.2BL and δ m ) or full CTOD R-curve. The fracture toughness data are obtained from tests conducted to BS 7448 Parts 2 and 4 on a girth-welded 14in ODx14.3mm WT CMn pipe using full-thickness square BxB SENB specimens (B = W =13mm). The programme comprised three nominally similar specimens notched into the weld metal (WM) and another set of three nominally similar specimens notched into the heat affected zone (HAZ). All six specimens were surface-notched toa crack/width ratio (a 0 /W) of 0.55, side-grooved and tested using the unloading compliance technique at 0°C. The HAZ R-curves had approximately double the slope of the WM R-curves, ie the toughness implied by the former for a given crack extension was approximately twice as large as the toughness implied by the latter. Attention here is focused on results obtained from one of the WM tests, which are broadly similar to results from the other two nominally similar specimens and are illustrated in Figure 6.
Fig.6. Single-point CTOD and CTOD R-curve considered
The following analyses are conducted using the Level 2B material-specific continuous yielding FAD that was adopted in the case study on choice of FAD:
- Level 2B analyses using two estimates of single-point fracture toughness: δ 0.2BL (= 0.19mm) and δ m (= 0.24mm corresponding to Δa m = 0.4mm).
- Level 3B analyses using the CTOD R-curve (shown in Figure 6) with a maximum allowable tearing, Δa max , ranging from 0.4mm to 1.0mm.
Results from the Level 3B analyses are shown in Figure 7 in terms of the ratio of maximum tolerable installation stress to SMYS vs. maximum tearing ( Δa max ). Results from the Level 2B analyses are also shown, but since these do not consider tearing, they are illustrated as horizontal lines denoted as ' δ 0.2BL Level 2B' and ' δ m Level 2B' (ie maximum tolerable installation stress is independent of Δa max ).
Fig.7. Results illustrating effects of fracture toughness on maximum allowable installation stress vs. maximum tearing
Results from the Level 3B calculations indicate clearly the potential advantages of a tearing analysis: as Δa max is increased from 0.4 to 1.0mm, the maximum tolerable installation stress increases by approximately 6%. In addition, the flaw size after tearing is known, which makes it possible to consider the stability ofthe flaw under subsequent loading. By contrast, results from the Level 2B analyses are simpler to perform but allow no or very limited advantage to be taken of the materials tearing resistance. As noted above, analyses employing δm are based on the material's toughness at Δa m but since this cannot be determined from conventional single-point tests it cannot be incorporated explicitly into the analysis. Consequently it is not possible to determine the flaw size accurately at the end of a Level 2 analysis and a small error will be introduced if the same flaw needs to be assessed under subsequent loading. Since Δa m is known in the present analyses, the error due to using a δ m result can be determined from Figure 7 as follows: the horizontal line denoted ' δ m Level 2B' crosses the line denoted 'CTOD ( δ) R-curve Level 3B' at Δa max = 0.5mm. This implies that the flaw size at the end of the δm Level 2B analysis should be increased by 0.5mm (in this particular case). Alternatively, the error can be expressed in terms of the maximum tolerable installation stress, which is 2% higher when obtained from a δm Level 2B analysis than from a Level 3B analysis (if the maximum allowable tearing in the latter is Δa m , ie equal to tearing at δm).
It should be noted that the above findings reflect the specific parameters considered in this case study and in particular the fracture toughness data used. As noted earlier, the magnitude of the error arising from a δ m Level 2B analysis depends on the ratios Δa m /a and δ m / δ 0.2BL . Although the effects of such errors are likely to be compensated for by other aspects of the analysis, it is recommended to conduct sensitivity analyses to establish whether results from δ m Level 2B assessments are significantly influenced by small variations in the fracture toughness and flaw size. In general, it is recommended to either use initiation fracture toughness values (ie δ 0.2BL or J 0.2BL ) in Level 2 analyses or, where necessary, use R-curves in Level 3 analyses to enable the flaw size to be updated (by the addition of stable tearing occurring under each load condition).
The above case study considers fracture toughness data determined from deeply cracked SENB specimens, which are tested in accordance with standard procedures (such as BS 7448:Parts 2 and 4). The results obtained generally provide a lower bound estimate of fracture toughness. However, recent studies indicate that single edge notch tension (SENT) specimens can provide less conservative and more realistic estimates of the fracture toughness of pipe girth welds than standardised, deeply cracked SENB specimens. The difference in behaviour has been attributed to higher constraint (or tri-axial stress) in SENB specimens (loaded in bending) than in both SENT specimens (loaded in tension) and the pipe(predominantly loaded in tension). As a result of a joint industry Project conducted jointly by SINTEF, TWI and DNV and the publication of its findings as a DNV Recommended Practice ( DNV 2006), SENT specimens are increasingly used in the offshore pipeline industry to determine fracture toughness for assessing pipelines subjected to significant plastic straining. However, further work is needed to determine whether and how SENT specimens can be used to determine safe estimates of fracture toughness under other conditions, eg when the pipe girth weld has a low toughness and where it is subjected to elastic or biaxial stresses.
Welding residual stresses
Attention is focused on assessing the options available for estimating the magnitude and distribution of welding residual stresses transverse to circumferential girth weld flaws. BS 7910 requires that these are included in fracture assessments and provides guidance covering welds in the as-welded and post-weld-heat-treated conditions, including models to allow for relaxation of residual stresses due to proof loading and/or interaction with applied (mechanical)stresses.
The simplest option is to assume that the welding residual stress is constant through the pipe wall and has a magnitude equal to the yield strength of the weaker of the pipe and weld metals. This initial stress is then allowed to relax according to Clause 22.214.171.124 of BS 7910, which essentially states that the relaxation is proportional to L r σ y / σ f , where L r is the plastic collapse parameter, and σ y / σ f is the ratio of yield to flow strength (= 1/L r max , where L r max is the maximum permitted limit of L r ). The maximum relaxation occurs when L r = L r max , and this leaves a minimum relaxed residual welding residual stress of 0.4 σ y . This model applies only if the residual stress is initially assumed to be of yield magnitude.
Another option is to use through-thickness residual stress distributions given in Clause Q.3.2 for a range of welding heat input. These are upper-bound profiles, which are not necessarily self-equilibrating, and are largely based on experimental data. The distributions are given for three levels of E 1 /B, where E 1 is the electrical energy per unit length of the largest weld run and B is the pipe wall thickness, namely low heat input (E 1 /B ≤ 50 J/mm 2 ); medium heat input (50 < E 1 /B ≤ 120 J/mm 2 ); and high heat input (E 1 /B > 120 J/mm2). The abovementioned relaxation model does not apply to these distributions, which in some cases limits their potential benefit.
A third option is to use data from other sources including experimental and/or numerical results determined for the component under consideration. This potentially provides the most accurate assessment but can be both costly and time-consuming to implement.
A case study is performed aimed at examining the first two options. This consists of three analyses, which are similar to the Base Case except that the distributions described above for the low, medium and high heat input are used to define the welding residual stress. The distributions are shown in Figure 8. This indicates that all three profiles give residual stress in the weld root equal to σ y (yield strength), which decays rapidly in the adjacent third of the thickness. The profiles are widely different in the two thirds of the thickness adjacent to the weld cap. The residual stress at the cap is equal to σ y , 0.2 σ y and -0.4 σ y respectively in the case of the low, medium and high heat input distributions. Linearised distributions, which enable membrane and bending stresses to be determined for each of the three heat inputs, are shown in Figure 9 superimposed onto the original non-linear distributions. The linearised distributions are determined such that the resulting stress is greater than or equal to the original non-linear stress at least over a 3mm wide area of the thickness, which contains the 3mm flaw in the cap region. The linearised distributions are converted into membrane and through-wall bending components (Q m and Q b , respectively) as follows:
- Low heat input: Q m = 448MPa, Q b = 0. This can be used to assess any flaws since the linearised stress lies above the non-linear distribution along the whole thickness. Two analyses are performed: one with residual stress relaxation according to Clause 126.96.36.199 (which is valid since the residual stress is of yield magnitude and constant through the wall thickness), and one without residual stress relaxation.
- Medium heat input: Q m = 225MPa, Q b = -135.4. This is applicable to flaws with a height ≤ 3mm and is performed without residual stress relaxation.
- High heat input: Q m = 0, Q b =0. This is applicable to flaws with a height ≥ 5.7mm (5.7mm is the width of the thickness region in which the original non-linear distribution is compressive).
Results from the analyses employing the above distributions are shown in terms of minimum required CTOD vs. P m /SMYS in Figure 10. The results reflect clearly the residual stresses assumed for each of the three heat inputs. The highest and lowest required CTODs are obtained, respectively, from the analyses using the most tensile linearised residual stress (low heat input) and the least tensile linearised residual stress (high heat input). The two sets of results for the low heat input case illustrate clearly the substantial benefits obtained from applying the relaxation model.
Fig.8. Residual stress distributions for three levels of welding heat input (according to BS 7910 Annex Q)
Fig.9. Linearised residual stress distributions of profiles shown in Figure 8
Fig.10. Results illustrating effects of assumed residual stress distributions on minimum CTOD (required to tolerate a 3x50mm flaw in the weld cap)
An examination of the original distributions in Figure 8 shows that if the 3mm flaw is located in the root, the most and least tensile linearised residual stresses will be those of the high and low heat inputs, respectively. This is the complete opposite to the linearised distributions obtained above for the cap flaw. However, since the differences between the actual distributions in the root are less pronounced than in the cap, differences between results for the root flaw (in terms of minimum required CTOD vs. P m /SMYS) will be less significant than the differences observed above for the cap flaw. It should be noted that linearisation of the stresses for the root flaw is not straightforward since the stress changes rapidly in the root region. Clearly, other options to determine stress intensity factor solutions for non linear residual stress distributions, such as weight functions (which preclude the need for stress linearisation), are highly desirable.
The above case study shows that application of the residual stress profiles has a significant effect on the assessment of flaws located in the cap region. However, for root flaws, differences between the distributions are less significant and linearisation of the profiles is more difficult than for cap flaws, due to the rapid decay in stresses. The residual stress relaxation model, applicable only when the stress is of yield magnitude and uniform through the thickness, produces significantly less conservative results than those obtained without relaxation. An assessment strategy would be to use the method that gives the most favourable results since all methods considered above are valid.
Summary and conclusions
Four case studies on assessing the significance of circumferential girth weld flaws according to the fracture assessment procedures of BS 7910 have been performed. The case studies consist of estimating the minimum CTOD required to tolerate a 3x50mm surface fusion boundary flaw on the outer surface of a typical offshore pipe girth weld under the effect of installation loading, or alternatively estimating the maximum installation stress that can be tolerated assuming a range of single-point fracture toughness values or a full R-curve. The objective of the case studies is to demonstrate the effects of selection of the FAD, tensile properties, fracture toughness properties, and welding residual stresses. Key findings are as follows:
- The material-specific Level 2B FAD provides less conservative and significantly more favourable results (expressed in terms of minimum required CTOD) than the generalised FAD for both continuous and discontinuous yielding. A comparison of results obtained using material-specific Level 2B FADs shows that, for L r < 1.0, more favourable results are obtained from the discontinuous yielding FAD than the continuous yielding FAD. The opposite applies for L r > 1.0. Recommendations have been given on selection of stress-strain curves for construction of material-specific FADs.
- Using actual tensile properties rather than specified minimum values enables more favourable results (expressed in terms of minimum required CTOD) to be obtained. Recommendations have been given on selection of actual tensile properties.
- A level 3B tearing analysis offers significant advantages over a single-point Level 2B analysis. These include more favourable results (in terms of maximum allowable installation stress) and the ability to determine the flaw size after tearing in comparison with a Level 2B analysis using a ?m value. The latter analysis includes an error when the initial flaw dimensions are used (ie the flaw size before application of load) because the toughness used represents the material resistance after a certain amount of tearing. The case study, considering specific fracture toughness data, shows that the flaw size at the end of the δm Level 2B analysis should be increased by 0.5mm or alternatively the maximum allowable stress should be decreased by 2%.
- The application of the residual stress profiles recommended in Annex Q of BS 7910 has a significant effect on the assessment of flaws located in the cap region. The most and least tensile residual stresses are obtained from the low and high heat input distributions, respectively. The opposite is true for root flaws, where differences between the distributions are less significant and linearisation of the profiles is more difficult due to the rapid decay in stresses in the root region. The residual stress relaxation model, applicable only when it is assumed that the initial stress is of yield magnitude and uniform through the thickness, produces considerably less conservatism results than those obtained without relaxation.
- API, 1999: American Petroleum Institute (API) Standard 1104, 'Welding of pipelines and related facilities', API, Nineteenth Edition, 1999.
- BEGL, 2006: 'Assessment of the integrity of structures containing defects', R6 Revision 4 and amendments', British Energy Generation Ltd, 2006.
- BSI, 1980: BS PD 6493:1980, 'Guidance on some methods for the derivation of acceptance levels for defects in fusion welded joints', British Standards Institution, 1980.
- BSI, 1991: BS PD 6493:1991, 'Guidance on methods for assessing the acceptability of flaws in fusion welded structures', British Standards Institution, 1991.
- BSI, 1997a: BS 7448 Part 2:1997: 'Fracture mechanics toughness tests, method for determination of K Ic , critical CTOD and critical J values of welds in metallic materials'. British Standards Institution, 1997.
- BSI, 1997b: BS 7448 Part 4:1997: 'Fracture mechanics toughness tests, method for determination of fracture resistance curves and initiation values for stable crack extension in metallic materials'.
British Standards Institution, 1997.
- BSI, 2000: BS 7910:1999 (incorporating Amendment 1), 'Guide on methods for assessing the acceptability of flaws in metallic structures', British Standards Institution, 2000.
- BSI, 2005: BS 7910:2005, 'Guide for assessing the significance of flaws in metallic structures', British Standards Institution, 2005.
- DNV, 2006: DNV: 'Fracture control for pipeline installation methods introducing cyclic plastic strain'. Document DNV RP F108, 2006.
- Knauf G and Hopkins P (1996), 'The EPRG guidelines on the assessment of defects in transmission pipeline girth welds', 3R International, 35, 10-11/1996, S. 620-624, 1996.