Henryk G Pisarski, TWI Ltd, Granta Park, Great Abington, Cambridge, CB1 6AL, UK
Colin M Wignall, TWI Ltd, Granta Park, Great Abington, Cambridge, CB1 6AL, UK
Proceedings of IPC 2002 International Pipeline Conference
September 29 - October 3 2002, Calgary, Alberta, Canada
The relationship between fracture toughness estimated using standard single edge notch bend (SENB), single edge notch tension (SENT) test specimens and fracture toughness associated with a circumferential flaw in a pipe girth weld is explored in terms of constraint using the Q parameter. It is shown that in the elastic-plastic regime, use of standard deeply notched SENB specimens provides a conservative assessment of fracture toughness, for both weld metal and HAZ, because of the high constraint associated with this specimen geometry. Use of specimen geometries and loading modes associated with lower constraint (e.g. SENT and shallowed notched SENB specimens), allow for improved estimates of fracture toughness to be made that are appropriate for the assessment of circumferential flaws in pipe girth welds. Recommendations are given on the specimen designs and notch orientations to be employed when evaluating weld metal and HAZ fracture toughness.
In order to provide quantitative assessment of the significance of fabrication flaws in pipeline girth welds, current practice is to carry out fracture mechanics analyses employing results from fracture mechanics tests obtained from deeply notched bend specimens. These tests are conducted in accordance with standard procedures such as BS 7448, ASTM 1290 or 1820 to derive fracture toughness in terms of crack tip opening displacement (CTOD) or J. The results obtained generally provide a lower bound estimate of fracture toughness which enables a conservative assessment of a known or hypothetical flaw to be made. However, the performance of large-scale pipe bend tests on pipeline girth welds has shown that such assessments can be overly conservative, especially when the fracture mechanics specimen fails in an essentially brittle manner whilst the pipe girth weld survives large plastic strains without unstable fracture. The difference in behaviour has been attributed to differences in crack tip constraint between the fracture toughness specimen and a flaw in the pipe, with constraint being higher in the former.
The need to resolve this problem has become acute because of the increasing need to demonstrate the integrity of pipeline girth welds when subjected to high plastic straining during installation (e.g. by reeling methods) or in-service (e.g. subsidence, earthquake or pipeline movement due to other environmental effects). Until recently it has been difficult to quantify constraint. However, elastic-plastic finite element analysis now permits constraint to be effectively modelled using the Q parameter [4-5] . This not only enables the effect of differences in specimen geometry and crack size to be assessed, but also the effects of strength mismatch between the parent pipe, heat affected zone (HAZ) and weld metal. The Q parameter, or constraint, is expressed as the normalised difference in the crack tip stress field in the test geometry ( σ TG) and a reference, plane strain small-scale yielding ( σ SSY) stress field for a crack in an infinite plate at a specified distance from the crack tip, as illustrated in Fig.1. When the crack tip stress field in the test geometry is similar to or exceeds that for the reference, small-scale yielding solution, Q is close to zero or positive and constraint is high in the test geometry. On the other hand, when the crack tip stress field is below that for small-scale yielding, Q is negative and constraint is low in the test geometry. It may be noted that Q is not a constant for a given test geometry but is a function of loading because the crack tip stress fields will be modified by the combined effects of work hardening mode of loading and shape of the test specimen or structural component. This model assumes that materials in the test and structural component geometries are homogeneous and the same. Since welds are inhomogeneous it is necessary to ensure that the fracture toughness specimen is designed to test all the microstructures likely to be encountered by flaws possible in the pipe girth weld.
Fig.1. Illustration of constraint parameter, Q, for a given applied J
Modelling of constraint
In the following, constraint was assessed as a function J derived from elastic-plastic analyses using ABAQUS. The specific girth weld considered was in a 324mm OD x 25.4mm WT pipe to API 5L X52. A shielded metal arc girth weld was made in the 5G position in a Vee bevel with 60° included angle. The average yield strength of the parent pipe was 440MPa (tensile strength 530MPa), whilst the (0.2% proof) average yield strength (0.2% proof) of the weld metal was 515MPa (tensile strength 605MPa).
Surface breaking flaws 50mm long, were inserted into the pipe model which was loaded in tension and bending up to 3.5% plastic strain. Most of the flaws were located in the centre of the weld on the outside diameter of the pipe or weld cap. Others were located in the weld root and fusion boundary. (Welding flaws, should they arise, are likely to be located at these positions). J and Q were then determined at various strain levels.
In addition, J and Q values were established at different load levels for square section single edge notch bend (SENB) and tension specimens (SENT). These represent specimens taken transverse to girth weld with a cross section of 25.4mm (B) x 25.4mm (W) and notched in the through-thickness direction to a/W = 0.13 and 0.5.
Constraint in test specimens
compares the constraint (assessed by the Q parameter) in the SENT and SENB specimens where the notches are located in the weld metal. As load increases (i.e. an increase in J), constraint decreases. (In the figures, J has been non-dimensionalised by the material yield strength and remaining ligament beneath the notch or flaw, b.) With the deeply notched specimens (a/W = 0.5), there are large differences in constraint between the SENT and SENB specimens (except at low loads where linear elastic conditions pertain) with the latter exhibiting higher constraint than the former. The differences in constraint manifest themselves as differences in fracture toughness as illustrated by the CTOD resistance curves for the two specimen geometries shown in Fig.3
. (These tests were conducted at room temperature so that brittle fracture was avoided). The specimens were prepared from SMAW girth welds made using E8010-G electrodes for the root and hot pass and E9018-G electrodes for the fill and cap. The resistance curves were obtained from over square specimens - 2BxB, where B or the W dimension is the pipe wall thickness which, in this case, was 20.6mm. The specimens were notched to a/B of 0.25 from the weld cap. Both SENT and SENB specimens were side grooved to a depth of 10% thickness; this increases constraint.
Fig.2. Constraint in SENT and SENB specimens for notch located in centre of weld
Fig.3. Comparison of weld metal CTOD R-curves from SENT and SENB specimens (2BXB, a/W=0.28, side grooved)
The effects of constraint are also apparent at lower temperatures in the transition regime where cleavage occurs. This is illustrated by a set of data from a different weld in which fracture toughness is described in terms of CTOD, see Fig.4. These welds were made using the SMAW process with electrodes to E8018-C1. Results from both shallow (a/W = 0.26) and deeply notched (a/W = 0.5) SENB specimens give higher transition temperatures than tests conducted on shallow notched (a/W = 0.28) SENT specimens. However, at low temperatures (in this case -50°C), differences in constraint and hence specimen geometry are minimal, so fracture toughness values are the same. Consequently, a single parameter can be used to define fracture toughness, in this case CTOD.
Fig.4. Comparison of fracture behaviour of SENT and SENB specimens (notched into weld metal, joint 50mm thick)
Figure 2 shows that when shallow notches are employed (without side grooving) with a/W = 0.13, similar levels of constraint are achieved in both SENT and SENB specimens.
Although analyses to assess constraint in SENB and SENT specimens notched into the fusion boundary/HAZ were not conducted, a separate study by TWI examined the effect together with weld strength mismatch in a welded BS 7191 Grade 450 EMZ steel. Karstensen et al  included in their model the HAZ and the crack was located in the middle of the grain coarsened HAZ (GCHAZ) close to the fusion boundary. The strength of the HAZ always overmatched that of the weld metal and parent plate. Constraint was found to increase with increases in weld strength mismatch, relative to the parent plate, in the order: undermatching, even matching and overmatching. However, more significantly, comparison of constraint in deeply notched SENB and SENT specimens (a/W = 0.5) notched into the GCHAZ confirmed that it was always higher in the former specimen type. These conclusions are supported by the current study where the notches were located in the weld metal centre line.
Constraint in pipe geometry
Figures 5 and 6
show how constraint varies with J in a pipe containing a shallow and deep external cracks (50mm long) in the centre of the girth weld. Loading in tension results in slightly higher constraint than loading in bending, but this is only apparent with the shallow crack (a/t = 0.13). Here 't' represents the pipe wall thickness and can be considered to be the W dimension in surface notched fracture toughness specimens, or B dimensions in the through-thickness notched specimens; this is illustrated in Fig.7
. Except at very low J values, the shallow notched SENT and SENB specimens are subjected to higher constraint than the shallow notched pipe. Similar behaviour is shown by the comparison of the deeply notched SENB and SENT specimens with the deeply notched pipe (a/t = a/W = 0.5), although in this case constraint in the SENB and SENT specimens differ.
Fig.5. Constraint associated with shallow flaws on weld centre line in pipe loaded in tension and bending, and SENT and SENB specimens
Fig.6. Constraint associated with deep flaws on weld centre line in pipe loaded in tension and bending, and SENT and SENB specimens
Fig.7. Flaw/notch orientations in pipe/test specimens (SN - surface notched; TTN - through thickness notch)
It is clear from these comparisons that use of deeply notched SENB specimens imposes a significantly higher degree of constraint than experienced by pipe containing both shallow and deep flaws. This will result in an under-estimate of the fracture toughness that would be experienced by the flaw in the pipe. Closer constraint matching will be provided by shallow notched SENB and SENT specimens and result in a more appropriate estimate of the fracture toughness to be made for flaws in pipe girth welds.
So far, only flaws located in the centre of the weld have been considered. However, in pipe girth welds a more likely flaw is lack of side wall fusion, where one side of the flaw is located adjacent to the HAZ and the other in weld metal. In this situation, strength mismatch between the parent pipe and weld metal could also influence constraint.
Figure 8 shows that there is little difference in constraint between a central weld metal flaw located at the weld cap or at the weld root, despite the width of the weld be less in the former compared with the latter. However, if the flaw position is moved from the weld centre line to the fusion boundary on the external surface (weld cap region), there is a significant increase in constraint. (Note, in this case, the weld metal overmatches the strength of the parent pipe). Thus, even if the weld metal and HAZ were to have similar 'inherent' fracture toughness, fracture would be more of a risk from flaws located at the fusion boundary than within the weld metal.
Fig.8. Effect of flaw location in pipe girth weld on constraint
Choice of specimen geometry
Use of SENT rather than SENB specimens to evaluate fracture toughness appears to have an advantage because the mode of loading simulates tension loading experienced by the wall of the pipe under global bending or tension. Analyses of constraint confirm that for deeply notched specimens (a/W = 0.5) with the notch located in either GCHAZ or weld metal, constraint is lower in the SENT compared with the SENB specimen and is closer to the constraint experienced by a flaw located in the pipe girth weld, weld metal. However, for shallow notched specimens (a/W = 0.13) the difference in constraint is minimal, so either specimen type can be employed. To ensure that constraint is not underestimated, it is recommended that fracture toughness is evaluated using either SENT and SENB specimens containing notch depths greater than the deepest/highest flaw expected in girth weld. However, the disadvantage of using the SENT specimen is that currently there are no standardised testing and fracture toughness evaluation procedures, so results may be affected by the particular procedures employed by the testing laboratory (i.e. there could be laboratory to laboratory variations). In addition, fracture toughness estimation is best derived from finite element analyses which will allow for notch depth, work hardening and strength mismatch effects. However, this is not a convenient or practical approach for most pipeline projects. A joint DNV-TWI-SINTEF project is working on a guideline document concerned with the assessment of pipe girth welds subjected to repeated plastic straining 
. This will include equations for estimating J from SENT specimens.
The advantage of using SENB specimens is that the specimen preparation and testing procedures are fully described in standards such as BS 7448 and ASTM 1820 and 1290. However, a draw-back is that the equations for calculating fracture toughness are only valid for deeply notched specimens (0.45 ≤ a/W ≤ 0.7). Errors in fracture toughness estimation can arise if the equations are applied to shallow notched specimens, in this case a/W < 0.45. This problem can be overcome if fracture toughness, J, is estimated from crack mouth opening displacement (CMOD). Kirk and Dodds  showed that J can be estimated for a wide range of a/W values and material strain hardening from CMOD using the following equation:
The stress intensity factor, K, at maximum force, is employed to estimate the small-scale yielding component of J, whilst the plastic component of the area under the force versus CMOD curve (A c) is used to estimate the plastic component of J.
The equation is valid for 0.05 ≤ a/W ≤ 0.7. CMOD can be derived directly from clip gauge opening when integral knife edges are employed on the specimen. Alternatively, a dual clip gauge arrangement can be employed such that knife edges are attached at the notch mouth. (The attachment is normally made by laser or micro TIG welding). CMOD is then estimated from:
V 1 and V 2 are lower and upper clip gauge openings, respectively
Z 1 and Z 2 are lower and upper knife edge heights, respectively
The equation for estimating J has been incorporated into DNV OS F101  and is being considered in a draft annex to ASTM 1290 on weld metal fracture toughness testing  .
Ideally, the notch orientation employed in the SENT or SENB specimen should represent that of the flaw to be assessed in the girth weld. Furthermore, to ensure that constraint is not less than that in the pipe, the notch depth employed should be greater than the deepest/highest flaw that could be present in the pipe. Normally, this would require use of surface notched specimens, notched from either the weld root or weld cap. However, unless it is known that the material is on the upper shelf at the test temperature (i.e. brittle fracture will not occur), there is a possibility that any local brittle zones present within the HAZ or weld metal may be missed by surface notching. This risk can be minimised by employing a through-thickness notch so that a wide range of microstructures (weld beads and associated HAZs) are sampled by the crack tip. Local brittle zones present in the weld will be intersected by the fatigue crack tip. However, testing the HAZ in this way can be problematic if a wide bevel angle is employed to make the girth weld, because the crack tip will only sample a small part of the HAZ; most of the crack tip will be in parent pipe and weld metal. The problem can be minimised by testing sets of specimens where the crack tip in each set is arranged to intersect a different position along the weld fusion boundary, for example, at the ¼, ½ and ¾ pipe wall thickness positions. This ensures that a range of HAZ microstructures are tested. The problem is less acute when the girth weld is made using a mechanised GMAW process where steep bevel angles are usually employed. Here, the near vertical fusion boundary ensures that with one set of tests most of the crack tip in the through-thickness notched specimen can be located in the HAZ.
From considerations of crack tip constraint, in terms of the Q parameter, in fracture toughness specimens and girth welds in pipes containing welding flaws, the following conclusions are drawn:
- Standard deeply notched SENB specimens will provide a conservative assessment of fracture toughness compared with a circumferential flaw in pipe subjected to axial or bending loading.
- An improved estimate of fracture toughness appropriate for the assessment of circumferential flaws in pipe girth welds can be provided by testing SENT or shallow notched SENB specimens.
- In the absence of standardised SENT testing procedures, use of shallow notch SENB specimens prepared and testing according to standard procedures should be considered as an alternative.
- It is recommended that the notch depth employed (for both SENB and SENT specimens) should exceed the size of the flaw considered to be present in the pipe girth weld. Furthermore, to ensure that any potential local brittle zones in the HAZ or weld metal are evaluated, it is recommended that the through-thickness notch orientation is employed. Modifications to the standard fracture toughness estimation procedures are recommended when employing shallow notched SENB specimens (when a/W < 0.45).
- DNV-TWI-SINTEF Joint Industry Project on 'Fracture control for installation methods introducing cyclic plastic straining. Development of guidelines for reeling pipe' (TWI Project 12201), to be published2002.
- DNV Offshore Standard OS-F101, 2000: 'Submarine pipeline systems', Det Norske Veritas, Hovik, Norway.
- Kirk, M.T., and Dodds, 1993, 'J and CTOD estimation equations for shallow cracks in single edge notch bend specimens'. Journal of Testing and Evaluation, TJT EVA, Vol.21, No.4, 228-238.
- O'Dowd, N.P., and Shih, C.F., 1991, 'Family of crack-tip fields characterised by a triaxiality parameter: Part 1 - Structure of applications', Journal of Mechanics and Physics of Solids, 40, 989-1015.
- O'Dowd, N.P., and Shih, C.F., 1992, 'Family of crack-tip fields characterised by a triaxiality parameter: Part 2 - Fracture applications', Journal of Mechanics and Physics of Solids, 40, 939-963.
- Karstensen, A.D., Horn, A., and Goldthorpe, M., 2002, 'Constraint loss in welds due to geometry, loading mode and strength mismatch', 2 nd International Symposium on High Strength Steel, Stiklestad, SINTEF, Norway.
- Wang, Y.Y., Reemsynder, H.S., and Kirk, M.T., 1997, 'Interference equations for fracture toughness testing: Numerical analysis and experimental verification'. ASTM STP 1321.