TWI Ltd
Anthony Horn
AMEC
Paper presented at the 6th International Pipeline Technology Conference (Rudi's Pipeline Conference), 6-9 October 2013, Ostend, Belgium.
Abstract
A British Standard (BS 8571) to determine fracture toughness using single edge notch tension (SENT or SE(T)) specimens in steel is currently being prepared. This is being done to reflect the BS working committee’s desire to standardise current practices and capitalise on over 8 years’ experience gained, mainly from the offshore pipeline industry, in using this test method. The standard will be primarily based on the method described in DNV RP F108, modified to reflect updated practises and recommendations from current literature, while referring to BS EN ISO 15653 for methods and validation of weld and HAZ specimens. Although there has been much literature published on SENT testing, there are still sizeable gaps in the justification of various factors needed for full standardisation. In this paper the validation of the determination of CTOD for SENT specimens, the acceptable limits of pre-crack curvature, and the tolerance on clip gauge position are all discussed. Areas where further work is required are also highlighted.
Introduction
A British Standard (BS 8571) to determine fracture toughness using single edge notch tension (SENT or SE(T) specimens in steel is currently being prepared. This will be largely based on the method described in DNV RP F108. This is to reflect the BS working committee’s desire to standardise current practices and capitalise on over 6 years’ experience gained, mainly from the offshore pipeline industry, in using this test method (Pisarski, (2010). The main purpose of the test has been to determine fracture toughness, in terms of J-integral or CTOD (Crack-Tip Opening Displacement), in ductile steels (where cleavage fracture does not occur) using a specimen design that is representative of the crack-tip constraint conditions associated with circumferential weld flaws in pipelines. The SENT test is used to generate a fracture toughness resistance curve (R-curve) by using a multiple specimen method or single specimen where crack extension is derived by either unloading compliance or potential drop methods. The results are typically used in conjunction with fracture mechanics analyses to set flaw acceptance criteria for pipelines that are subject to plastic straining during either installation or service. For the test to become standardised its use needs to be validated for both brittle and ductile materials and for single point fracture toughness as well as R-curves, and be applicable to a wider range of industries.
Existing Procedures
There have been a number of SENT test procedures described in the last decade. Researchers in Norway (mainly at SINTEF) have published research on SENT testing broadly based on the DNV RP F108 procedure. ExxonMobil have also developed their own procedure based on a large programme of research they have carried out internally (Tang, 2010). There are also a number of published papers on SENT testing by two main teams of researchers apart from TWI; there is an active Canadian group based at Canmet, and a number of Brazilian researchers based at universities and industrial companies around Sao Paulo. Each group has published their own unloading compliance methods for SENTs (Cravero & Ruggieri, 2007; Shen et al, 2009). Commercial test houses use their own methods based on DNV RP F108 but often follow the validity criteria of BS 7448 Part 1, ISO 12135 or BS EN ISO 15653 (for SENB specimens). The various sources of SENT test research from the literature has been collated and reviewed by the BS working committee for consideration for standard BS 8571. The independent research teams have evolved quite separate methods, and so the choice of different test parameters is currently quite broad, see Table 1.
Specimen design and preparation
SENT specimen cross section dimensions are expressed as the specimen thickness x width, BxW (where W is measured in the material thickness direction), see Figure 1. Within the new BS 8571 standard, the dimension B is equal to the W or 2W at the option of the customer. Test specimens have the dimension W which is as close as possible to the full thickness of the material after machining. When preparing specimens from pipe sections, the machining is required to be the minimum necessary to remove pipe curvature so that a rectangular cross section is obtained. The surfaces of the specimen are required to be fully machined to meet the tolerances and surface finish specified by ISO 12135. Furthermore, the specimen needs to be straight along its length and straightness tolerances are defined. This is necessary in order to prevent additional bending stresses being induced by the specimen straightening itself under initial loading during the test. The length of the specimen will depend on whether it is loaded by clamping the ends or pin loaded. For clamped specimen the daylight between the grips is defined as ten times width (W) with typically an additional 100mm length required at each end to enable fitting into the grips. The requirements for pin loaded specimens are yet to be finalised but is likely to involve clamping the ends and pin loading these grips. For pin loaded specimens the day-light between the grips need not be 10W.
Specimens are notched from the surface defined by the customer (original plate or pipe surface) to a final depth (after notching and fatigue precracking) in the range equivalent to 0.2<a/W<0.5. Although DNV RP F108 only describes surface notching, the BS 8571 standard will include an option to notch through thickness as well. Notches are positioned to locate the final crack tip in parent material, weld metal or heat affected zone (HAZ), as defined by the customer. Notching is carried out using electro-discharge machining (EDM) such that the notch width at the notch tip is no wider than 0.3mm. Specimen preparation procedures, including fatigue precracking are based on those given in ISO 12135 and BS EN ISO 15653, so fatigue precracking can be carried out more conventionally in three point bending. The latter standard provides guidance on how to locate notches in weld metal and HAZ. Typically, fatigue pre-cracking will be carried out in three-point bending to produce a fatigue precrack length of at least 2mm.
Side-grooving of the specimens sides after fatigue pre-cracking has been suggested as a means ensuring a straighter fatigue crack front (ie by minimising curvature) and helping to ensure that crack extension by ductile tearing during the tests takes place in a uniform manner along the whole of the crack front (ie it avoids the formation of a thumb nail shaped tear). Side grooving will tend to increase crack-tip constraint, which is counter to the objective of the test. However, since achieving a crack extension with a straight crack front is a desirable objective, it is being considered by the BS committee. ExxonMobil use a side groove of 5% each side (Tang et al) while Canmet favour 5 to 7.5% each side (Shen et al 2010) when using unloading compliance methods.
Testing welds
SENT testing has long been used for characterisation of girth welds in pipelines. The standard will make reference to BS EN ISO 15653 for guidance about fracture toughness testing of welds and HAZs, and the requirements for post-test metallography.
When shallow notches are located in weld metal that over-matches the yield strength of the parent material, there is a risk of preferential yielding of the parent material that forms the arms of the specimen resulting little opening of the crack. To avoid this, guidance will be provided on the choice of minimum crack depth to be employed for a given weld metal strength over-match. An example of the a 0 /W that should be exceeded to avoid preferential plastic straining in the arms of the specimen notched into weld metal as function of the yield strength mismatch ratio (defined as the ratio of weld metal to parent material yield strength) is shown in Figure 2. In this example the specimen cross-section is over square (thickness, B, is twice the specimen width, W). This curve has been obtained by calculating the limit load for a notched specimen made of weld metal and varying the notch depth so the limit load is less than that for an un-notched specimen made of parent material only.
Specimen instrumentation and testing
The method will enable either J-integral or CTOD or both to be determined from the test. If only J-integral is to be determined, the a single clip gauge can be employed to measure crack mouth opening displacement (CMOD) provided that integral knife edges are machined at the notch mouth. Otherwise a pair of clip gauges, one mounted above the other, are employed which are positioned between knife edges spot welded on the notched surface close to the notch mouth. Finite element analyses have been conducted to establish maximum distance that the spot welds can located from the notch mouth and still obtain reliable J-integral and CTOD results. The modelling shows that for a typical SENT specimen (a/W=0.2, H/W=10, B=W=25mm) plasticity does not reach region immediately behind notch and therefore the knife edge positioning should not adversely affect CMOD measurement as long as knife edges are attached to the specimen within the elastic region (see Figures 3 and 4). For a knife edge attached at 9mm from the crack mouth, CMOD would be overestimated by 2-3%, while if attached at 6mm or less, the measured CMOD would be within 1% of actual CMOD.
During the SENT test, the applied tension force and clip gauge opening is recorded, as is normal in fracture toughness testing. Generally, the purpose of the test will be to generate a fracture resistance curve. This can be achieved using the multiple specimen method where a set of six identical specimens are loaded and each unloaded at different clip gauge opening displacements creating varying amounts of ductile crack extension. From the analysis of these data the R-curve is generated. Depending on how the results will be used, either a mean line or a lower bound will be fitted to the data. Alternative methods of generating the R-curve include use of the partial unloading method where the change in unloading compliance is used to determine crack length and the electrical potential drop method. With both these methods three specimens are employed for each notch location in order to provide an estimate of material variability.
The standard will also provide guidance for the determination of a single point J or CTOD in the case where a full R-curve cannot be achieved, although the testing standard will not provide advice for how to use such data, for example, for a fitness-for-service assessment.
Analysis of test data
J-integral and CTOD are determined by partitioning the elastic and plastic components of J and CTOD. The elastic component is derived from the force at the start of unloading and associated elastic stress intensity factor (K). The plastic component of J is obtained from the area under force versus plastic component of CMOD curve. Unless integral knife edges have been employed, CMOD is determined by extrapolating the readings from the pair of clip gauges mounted on knife edges above the notch mouth back to the crack mouth. This equation is given below.
Vp1 and Vp2 are the plastic parts of the clip gauge displacements for knife heights of z1 and z2, respectively.
The equations for calculating J-integral are the same as those given in DNV RP F108. The basic form of these equations is given below:
Where,
Jel is the elastic component of J,
Jpl is the plastic component of J,
K is the elastic stress intensity factor at force applied to the specimen at the start of unloading
E’ is the longitudinal elastic modulus in plane strain, equal to E/(1-n2)
ηp is a dimensionless function of geometry,
Up is the plastic part of the load versus crack mouth opening displacement CMOD curve,
B is the specimen thickness,
W is the specimen width,
a 0 is the initial crack length (excluding crack extension due to tearing)
The elastic component of CTOD is determined from the elastic K and the plastic component of CTOD by extrapolating the readings from the pair of clip gauges back to the original fatigue precrack tip. The equation is given below.
Where σ y is the yield strength of the material in which the crack is located. When it is HAZ, the weld metal yield strength is used (as recommended in BS EN ISO 15653). ‘m’ is a constraint factor. Limited finite element analyses indicate it to be close to 1. However, if brittle fracture were to occur it might be necessary to establish a more precise value. However, when the SENT test is used to generate an R-curve the elastic component of CTOD is small and any errors in the assumption for ‘m’ have a negligible effect on the results.
Finite element analyses and a rubber infiltration method have been used to establish CTOD in SENT specimens. Although the number of cases examined is limited, they confirm that the above equation provides an acceptably accurate method of determining CTOD provided that a/W is equal to or exceeds 0.2, see Figure 5. If a/W<0.2 the error in estimating CTOD using the double clip gauge method is non-conservative and can exceed 10%. Consequently, when CTOD is to be determined, the standard will require a/W not to be less than 0.2, although the exact notch depth limits to be included in the standard have yet to be confirmed.
A validation of CTOD methods performed by TWI (Moore and Pisarski, 2012) showed that at a/W ratios of 0.3 to 0.5 an equation for calculating CTOD from J for SENT specimens given by Shen & Tyson (2009) offers the best alternative method to calculate CTOD from J compared to methods given in recent versions of the DNV OS F101 standard (Figure 6). A number of alternative methods for determining both CTOD and J-integral have been proposed in the literature; however, these have not yet been fully evaluated by the BSI committee. The method to determine J given in DNV RP F108 is well established, but a comparison with alternative J equations in the literature will be done as part of the validation for the standard. The advantage of the simple extrapolation method from the double clip described above is its simplicity to determine CTOD and has been extensively used by a number of laboratories and significant issues with the method have not been flagged.
Qualification requirements
The full range of test qualification requirements are yet to be agreed. These are likely to include limits on the amount of crack extension by tearing (DNV RP F108 currently limits this to 2 to 3mm and is likely to be retained), and the definition of a crack blunting (current standards for SENB tests define ductile crack extension as extension beyond stretch zone formation, as this is the creation of a fracture surface. According to DNV RP F108 the measurement of crack extension includes any stretch zone. Unless there is compelling evidence against this, the same definition will be employed in BS 8571
Some progress has been made in defining fatigue pre-crack front straightness requirements. The initial crack length, a0, is defined as in BS7448 Part 4 and is obtained by first averaging the two measurements at the outer points and then averaging this value with the seven inner points. The two outer measurements are taken 0.01B from the surface. Crack front curvature is defined as the difference between a0 and any of the nine crack length measurements. The maximum limit set by BS7448 Part 4 is 10%. A survey of fatigue pre-crack front curvature measured in over 400 SENT specimens is shown in Figure 7. If the curvature limit in BS 7448 Part 4 were adopted many of the results would be considered to be not qualified as the majority of the data give crack front curvature greater than 10%. However, does this apparently excessive curvature have an effect on the SENT test result? To assess this, the effect of fracture front curvature on J and CTOD was examined through a series of finite element analyses where the curvature was 4, 10 and 17% of a0 (Figure 8). The effect on the variation of CTOD and J across the crack front was examined and compared to a straight crack front. The SENT specimen had a cross-section with B=40mm and W=20mm, a/W was 0.33. The results are presented in Figures 9 and 10. They show that with up to 17% curvature, errors in J and CTOD, relative to a straight crack front, do not exceed 10% and 7%, respectively. Consequently, it is recommended that crack length measurements of SENT specimens are based on BS7448 Part 4 and that the maximum allowable curvature is 20% of a0, (Malpas et al, 2012).
Further Work
The structure of the British Standard for SENT testing, BS 8571 follows the flow chart given in Figure 11. The details of the validity limits still need to be decided by the BS working group, but the general method for SENT testing has been decided. The intention is that a draft for public comment will be available during 2013.
Further research work is required in the development of SENT testing to make the test type more general, a prerequisite for standardisation. TWI is running a Group Sponsored Project to help carry out this work in order to revise BS 8571 for the next version.
- Research will concentrate on validation of J and CTOD methods in weld and parent metal specimens.
- The specimen details will be confirmed (specimen dimensions, side grooves, notch orientation).
- Validating the use of SENT specimens for low temperature, brittle fracture and single point tests.
- Quantifying the effect of crack path deviation on the validity of the SENT test result.
- Providing guidance on the requirement or not of a blunting line for a SENT R-curve.
Conclusions & Recommendations
A British Standard on SENT testing (BS8571) is being drafted which builds on the experience gained in the offshore pipeline industry in using the guidelines in DNV RP F108. The procedure will enable both J-integral and CTOD resistance curves to be determined. Further work is necessary to define the full qualification requirements but some progress on this has already been made. It is expected that a draft for public comment will be available during 2013.
References
Malpas M, Moore PL & Pisarski HG (2012), “Crack front straightness qualification in SENT specimens”, ISOPE Conference, Rhodes.
Moore PL & Pisarski HG (2012), “Validation of methods to determine CTOD from SENT specimens” ISOPE Conference, Rhodes.
Pisarski H.G. (2010), “Determination of pipe girth weld fracture toughness using SENT specimens”, International Pipeline Conference, Calgary, Alberta, Canada, IPC2010-31123.
Shen G & Tyson WR (2009), “Evaluation of CTOD from J-integral for SE(T) specimens”, Pipeline Technology Conference, Ostend, 12-14 October.
Shen G, Gianetto J, Tyson W, 2009: ‘Measurement of J-R Curves using single specimen technique on clamped SE(T) specimens’, in Proceedings of the Nineteenth International Offshore and Polar Engineering Conference, ISOPE2009, Osaka, Japan, June 21-26.
Shen G, Tyson W, Gianetto J, 2010: ‘Effect of Side Grooves on Compliance, J-Integral and Constraint of a Clamped SE(T) Specimen’ in proceedings ASME 2010 Pressure Vessels and Piping Division/K-PVP Conference (PVP2010) July 18–22, 2010 , Bellevue, Washington, USA.
Tang H, Macia M, Minnaar K, Gioielli P, Kibey S and Fairchild D, 2010: ‘Development of the SENT test for strain-based design of welded pipelines’, In Proceedings IPC 2010 8th International Pipeline Conference, Calgary, Canada, 27 Sept.-1 Oct.2010.
Table 1 Summary of different published SENT testing methods.
|
DNV RP F108, Jan 2006
|
ExxonMobil, Tang et al, 2010
|
CanMet, Shen et al, 2009
|
Cravero Ruggieri, Eng Frac Mech 2007
|
Draft BS8571, 2013
|
Specimen Sidegrooved each side?
|
2BxB (recommended) No
|
BxB 5%
|
BxB 5 - 10%
|
Bx2B 10%
|
BxB or 2BxB 5% suggested
|
Notch location Orientation Precracked?
|
Parent or weld Surface OD Fatigue
|
Parent or weld Surface OD Fatigue or EDM
|
Parent metal
|
Parent metal
|
Parent or weld Surface or T-T Fatigue
|
a/W
|
0.2 – 0.5
|
0.25 – 0.35
|
0.25 and 0.5
|
0.5
|
0.3 – 0.5
|
Fracture toughness
|
J R-curve M/S
|
CTOD R-curve U/C or DCPD
|
J R-curve U/C
|
J R-curve U/C
|
J or CTOD Single point or R-curve M/S or U/C
|
Application
|
Installation of pipelines under high plastic strain
|
Strain capacity of pipeline girth welds
|
Pipe
|
Engineering structures under plastic regime
|
General
|
Welds
|
Weld notch Fusion line notch PTM
|
Parent metal Weld metal HAZ PTM
|
Parent metal only
|
Parent metal only
|
Parent metal Weld metal HAZ PTM
|
Validity
|
Not discussed. Segment tests?
|
Yes, range of validity limits
|
Based on E1820
|
No. Tests to verify FEA.
|
Based on ISO 12135 and 15653
|
Loading
|
Clamped or pin loaded
|
Clamped
|
Clamped
|
Clamped or pin loaded
|
Clamped or pin loaded
|
M/S = Multiple specimen
U/C = Unloading compliance
PTM = Post test metallography