Вы находитесь на странице: 1из 17

Engineering Fracture Mechanics 71 (2004) 19972013 www.elsevier.

com/locate/engfracmech

A specimen for studying the resistance to ductile crack propagation in pipes


A. Shterenlikht a, S.H. Hashemi a, I.C. Howard J.R. Yates a, R.M. Andrews b
a

a,*

Department of Mechanical Engineering, University of Sheeld, Mappin Street, Sheeld S1 3JD, UK b Advantica Technologies Ltd., Ashby Road, Loughborough, Leicestershire LE11 3GR, UK Received 27 January 2003; received in revised form 25 July 2003; accepted 1 October 2003

Abstract The paper describes a novel DCB-like specimen designed for studying the resistance of running axial cracks in pipelines. It is long enough for steady-state conditions to govern a large proportion of the crack growth, a condition monitored by direct observation of the crack tip opening angle (CTOA) using video cameras. To avoid possible inaccuracies due to pre-straining the material, realistic specimens must be cut from the pipe and tested without attening. The geometry of the specimen design allows this, and also includes a measure of gauge-length thinning. Extra thickness due to loading side-plates accentuates two desirable features of gauge thinning, by providing directional stability in the advancing crack tip, and by increasing the level of constraint. Tests on specimens from API X80 grade pipe demonstrated a common pattern of CTOA development for 4, 8 and 10 mm gauge thickness material. All showed a reduction of CTOA with growth towards a constant, steady-state value, with the steady state being associated with slant fracture. The region of reducing CTOA started with at fracture, and gradually developed increasingly denite slant fracture. The most accurate tests produced a steady-state mean value of CTOA of 11.2 and a standard deviation of 1. 2003 Elsevier Ltd. All rights reserved.
Keywords: Steels; Crack tip opening angle; Ductile fracture; Stable crack growth; Pipelines

1. Introduction There is a long standing and on-going discussion on the possibilities of using small specimens for the prediction of the fracture behaviour of industrial-sized structures. Some authors have claimed [1] that although the material characteristics of crack initiation (e.g. J -integral) can be measured on laboratory-sized specimens, crack propagation and arrest properties can only be measured by large-scale testing. Others simply reject the use of small-sized specimens because of the claim that the fracture toughness of a small specimen cannot safely predict the fracture behaviour of a large structure [2].
*

Corresponding author. Tel.: +44-114-222-7740; fax: +44-114-222-7890. E-mail address: i.howard@sheeld.ac.uk (I.C. Howard).

0013-7944/$ - see front matter 2003 Elsevier Ltd. All rights reserved. doi:10.1016/j.engfracmech.2003.10.001

1998

A. Shterenlikht et al. / Engineering Fracture Mechanics 71 (2004) 19972013

However other workers [36] have shown that successful safety and fracture propagation predictions for large industrial structures can be made using the experimental results obtained on small laboratory-sized specimens. This became possible with the development of local approach techniques. These damage mechanics methods assess the intrinsic fracture material properties through nite element modelling applied to specic laboratory data [6,7]. A major concern for the designers and operators of thin-walled pressurised cylinders such as pipelines or aircraft fuselages is the control of running ductile cracks. In a gas pipeline, these cracks propagate at speeds below the acoustic velocity in the gas and are driven by a complex interaction between the deforming pipe and escaping gas. A critical issue in controlling these fractures is the discovery of the parameters that characterise the behaviour of running ductile cracks in laboratory-sized specimens. This invariably involves a transition from a stationary crack to a steadily moving one. In thin structures under plane stress conditions, this transition can also involve a transition from at to slant fracture, the slant fracture being characteristic of the steady-state propagation conditions. It is this slant mode tearing that is most representative of running cracks in real pipelines. A number of authors [814] consider crack tip opening angle (CTOA) as a very promising and convenient fracture criterion, especially for running ductile cracks in pipes. A CTOA-based design criterion against crack propagation is usually written in the form [12,15]: CTOAmax < CTOA 1

where CTOAmax is a measure of the maximum crack driving force calculated from a knowledge of the dimensions, material properties and operating conditions of a pipe, and CTOA is the resistance of the material. The increasing evidence [12,14,16] that CTOA is a true material property supports the belief that the CTOA value measured on a small size specimen (laboratory) can be used in the analysis of a large industrial construction (pipeline). (The possibility of using CTOA in tuning of the nite element models of the corresponding experiments was recently studied [17,18].) The work reported in this paper has concentrated on the design of a laboratory-sized specimen aimed at reproducing the behaviour of an axial ductile crack in a pipe with the emphasis on CTOA measurement. The stressstrain eld ahead of a running crack tip in a pipe has several important features which are not easily reproducible in laboratory-sized specimens [11]. These are (1) The plastic zone in a pipe extends up to 2.5 pipe diameters ahead of the crack tip and about 0.3 diameters on each side of the crack line. (2) A region of high biaxial tensile strain and strain rate extends approximately one fth of the pipe diameter ahead of the crack tip in both longitudinal and circumferential directions. (3) There are positive longitudinal plastic strains. (4) Initial plastic tensile strains are longitudinal. High circumferential plastic strains can be observed only very close to the crack tip. Thus the ideal specimen for studying a running ductile longitudinal crack in a pipe has to be large enough to capture the extent of the plastic zone, it must promote positive minor strains and it has to be designed for a dynamic test. The CTOA measurement would be most accurate if the specimen were atsided (i.e. there were no side grooves) and if the ligament were long enough to achieve steady-state crack propagation conditions. However, in practice, dynamic CTOA measurement is very dicult [16]. For the sake of simplicity of CTOA measurement it was decided to concentrate on a quasi-static test. This of course leaves the dynamic eects out. The dierence between static and dynamic CTOA is still an open issue. Some workers have

A. Shterenlikht et al. / Engineering Fracture Mechanics 71 (2004) 19972013

1999

found it important to report the CTOA values according to the type of the test used, viz dynamic [12,16] or quasi-static [14]. According to this terminology, all CTOA values reported here are quasi-static. The literature contains a number of dierent specimen geometries for studying ductile crack propagation. The one introduced in [19] was used to investigate shear ductile fracture. It had a slant notch so that shear fracture would be observed from the very onset of crack propagation. Thus any transition from initial at to shear fracture was excluded from the test. Side grooves to a depth of 12.5% were machined on both sides to promote the initial inclination angle throughout the test. This specimen geometry and its triple pantleg modication were utilised extensively for mixed mode I/III (tear + antiplanar shear) and pure mode III tests [2024]. Although the above specimens were used successfully for the analysis of crack initiation they are hardly suitable for stable crack growth studies because of the short remaining ligament. Side grooves prevent the crack from propagating freely thus increasing the constraint in the test section. This may result in a smaller CTOA value than that measured on an unconstrained crack prole. The CTOA value measured on a side grooved specimen would not be a true material property but will also depend on the geometry of the grooves. Finally, the presence of side grooves makes direct CTOA measurement by optical means dicult. Shear crack arrest was investigated with a spring loaded double cantilever beam (DCB) specimen [25]. The specimen used was 1000 mm long and 440 mm wide plate with thickness varied from 18 to 32 mm. The specimen was side grooved with a groove radius of 50 mm leaving the test section thickness of just 1.46.4 mm [25]. This specimen is far too big to be cut directly from a pipe (because of wall thickness and pipe curvature). Its size is also impractical for a laboratory experiment. The test section is too thin to capture any variance in material properties through thickness. In linepipe this is primarily a microstructural variation, e.g. existence of bainitic bands close to the centreline, which can cause delaminations [13]. Similar spring-loaded tests were performed on 20 mm thick 250 mm long and 700 mm wide plates [26,27]. A 4000 kN dynamic tensile machine was employed and strengthening plates were welded around the loading holes in order to avoid failures. This specimen is also too big to cut from a pipe and requires a very large load capacity test machine that is not widely available. Additional welding makes it even less attractive for most purposes. The drop weight tear test (DWTT) [16] has long been used for dynamic crack propagation analysis. However the fact that the minor strains are inherently negative in DWTT test makes direct transfer of measured CTOA to pipe design problematic. A modied DWTT specimen has been proposed recently [13,14]. The authors introduced a back slot lled with a shim, ush with the surface. This changes the ligament length without changing the interaction mechanics of crack initiation and propagation. However, the curved DWTT specimens extracted from the pipe were straightened [13] to obtain at specimens. This operation inevitably causes additional plastic deformation and may change the fracture propagation behaviour. The above comment on negative minor strains is also true for the modied DWTT. Standard double cantilever beam (DCB) and tapered DCB specimens introduce bending as well as tension loads which is not desirable. However the basic DCB-like specimen geometry is convenient for laboratory tests. Thus a new DCB-like specimen is proposed. It exhibits the following characteristics: It may be cut directly from a pipe, without any attening. The maximum possible width, thickness and ligament provide a large plastic zone. The width and thickness are limited by pipe curvature and wall thickness. High constraint in the test section is promoted by two thicker loading arms. This serves two purposes. Firstly positive or at least non-negative longitudinal strains can be achieved. Secondly the loading is predominantly in tension with only a small shear component. The test section does not restrain the transition to slant mode shear fracture. The test section is at near the crack tip for ease of CTOA measurement.

2000

A. Shterenlikht et al. / Engineering Fracture Mechanics 71 (2004) 19972013

2. Experiments The tests were carried out on specimens cut from a trial length of X80 material pipe, 1219 mm diameter and 13.8 mm wall thickness. The material had a yield stress of 546 MPa and a tensile strength of 686 MPa, both values obtained on round bar tensile tests. The minimum specication values for these parameters are 551 and 620 MPa accordingly. The specimen geometry is shown in Fig. 1. It is an 11 mm thick plate with the test area machined down to 8 mm. The pipe wall thickness and pipe curvature limit the thickness and the width of the at plate that can be extracted from it. Loading of the specimen was achieved by clamping it between two pairs of substantial loading plates, one pair of which is shown in Fig. 2. The load-line is going through the rst pair of the loading holes and is 62 mm from the initial notch. Fig. 3 shows the loading arrangement at the end of the test. Loading was under displacement control of 0.1 mm/s in a Schenck 250 kN servo-hydrolic machine. Load and actuator displacement data were taken directly from the instrumentation of the machine. This machine is restricted to a maximum piston stroke of 100 mm, but this was enough to drive the crack stably all along the specimen. Since data on the motion of the crack tips and their anks came from optical records collected by video and still cameras, the surface of the gauge area needed to be suitably non-reective. A coat of dark matt blue paint sprayed onto the surface proved very suitable. Also, the surfaces needed grids imposed on them if their change of shape were to be re-created from optical images after the experiment. A square grid (see Fig. 6) of 1 mm spacing on one side of the specimen subsequently provided all the information needed for indirect estimates of CTOA. This came from individual frames captured at every 1 mm of crack growth and stored in the data-acquisition computer. Fig. 4 illustrates the data gathering scheme. The clock of the video camera was syncronised with that of the data acquisition computer. All images were taken with a 12 optical zoom and registered with a 720 576 resolution. Thus 1 mm of specimen

Fig. 1. Specimen design, showing two projections of the specimen and an enlarged view of the notch. All dimensions are in millimetres.

A. Shterenlikht et al. / Engineering Fracture Mechanics 71 (2004) 19972013

2001

Fig. 2. Auxiliary loading plates.

Fig. 3. A specimen during the test.

Fig. 4. Loading and data recording scheme, showing (a) digital video camera, (b) initial notch tip, (c) crack tip after fatigue precracking, (d) tip of the moving crack, (e) surface area recorded on video, (f) expected crack path, (D) measured grip displacement, (P) applied load, (PC) computer with data acquisition hardware and software.

2002

A. Shterenlikht et al. / Engineering Fracture Mechanics 71 (2004) 19972013

surface is represented by 35 pixels in the image. All CTOA and crack length measurements were performed with digital image measuring tools in GIMP software [28].

3. Results and discussion Fig. 5 shows a specimen after the test. Some of the images obtained on one of the specimens are shown in Fig. 6. The measured value of CTOA changes from initially high values (Fig. 6a) through lower, relatively constant values at the stable state (Fig. 6b and c) and nally rises to slightly higher values, associated with the plastic collapse at the end of the test (Fig. 6d). The variance of light intensity on the surface shown in Fig. 6 is associated with out-of-plane deformation of the free surface.

Fig. 5. A specimen after the test.

Fig. 6. Four of the captured images of the propagating crack showing (a) the crack tip at the beginning of crack propagation; the initial notch is also visible, (b) and (c) the crack tip at two instances during stable growth and (d) the plastic collapse stage of the test.

A. Shterenlikht et al. / Engineering Fracture Mechanics 71 (2004) 19972013

2003

Fig. 7. A direct CTOA measuring scheme.

3.1. Direct CTOA measurement The CTOA values were obtained using the images captured on video. Two techniques were used, the rst one being a direct angle measurement on the crack edges, as shown schematically in Fig. 7. In this, CTOA was measured by recalling an individual image recorded on video tape and (1) (2) (3) (4) locating the crack tip, locating points on both crack surfaces at some distance, x, behind the crack tip, tting straight lines between the crack tip and each point, and computing the angle, H, between the straight lines as H 2arctg y x 2

The near tip opening during steady-crack growth in this experiment varies through the thickness, and it is manifest on the surface patterns recorded optically during the test. Hence, a suitable measure of crack resistance must be a reection of the through-thickness average material performance if it is to be representative of the potential behaviour of a moving crack in a pipe-like wall. On this basis, the CTOA should have a scale length, x (Fig. 7) that is not too small to avoid picking up only local surface behaviour. Accordingly the value of 1 mm for x was used in the present work. Some authors [8] reported the values for x in the range 0.251.25 mm used for obtaining CTOA on 2.3 mm thick plates of aluminium alloys. Distances below 1 mm were unsuitable for the thicker plates of X80 material tested here because of large out-of-plane displacements in close proximity to the crack tip. 3.2. Indirect CTOA measurement The second technique involves measurement of an angle between the grid lines behind the crack tip. The technique is indirect since data from the crack edges are not used in this approach. A schematic illustration of the indirect technique is shown in Fig. 8. The important advantage of this technique is that the exact crack tip position, which sometimes is very dicult to determine from the captured video images, is not needed. Uncertainty in crack tip location inevitably causes scatter in CTOA values obtained by a direct technique (see the scheme illustrated in Fig. 7). This becomes particularly important if the crack path is not a straight one and the crack anks have a jagged appearance (Fig. 16a). In such cases a direct technique is not viable and the indirect one is the only useful CTOA measuring tool. This will be discussed in Section 4.

2004

A. Shterenlikht et al. / Engineering Fracture Mechanics 71 (2004) 19972013

Fig. 8. Scheme for indirect CTOA measurement.

3.3. CTOA results Fig. 9 shows the experimentally determined values of CTOA plotted against dimensionless crack length. About 80 images were analysed along the whole crack path. The CTOA values were measured for each image on the crack edges and on the rst, second and third pairs of grid lines counting from the crack anks. The maximum CTOA values are obtained at the very beginning of crack growth. These correspond to fracture mainly by at tearing. Initial CTOA values are very much aected by crack blunting and therefore their values are very sensitive to the distance from the crack tip. Thus the maximum initial CTOA, 45, is obtained from the rst pair of grid lines. The values obtained from the second and third pairs are significantly smaller, 26 and 17 respectively. Whereas the indirect technique gives results from the very onset of crack propagation, the direct technique can be applied only after some crack growth, about one half of the gauge thickness. The CTOA value obtained from the crack anks at this point is 23. The transition from pure mode I to mixed mode I + III (tear + antiplanar shear) is characterised by a decrease in CTOA values, since slant fracture requires less energy than at [29]. After a stable shear crack

45 40 35 30 CTOA, deg 25 20 15 10 5 0 0 1 Stable state Stable state values mean CTOA, deg STD, deg 3.1e-001 crack flanks 12.5 7.4e-001 1st grid line 11.1 6.9e-001 2nd grid line 11.2 6.4e-001 3rd grid line 11.3

crack flank 1st grid line 2nd grid line 3rd grid line

2 3 4 5 6 7 Crack length/specimen gauge thickness

Fig. 9. CTOAcrack length data. The crack length is measured from the notch.

A. Shterenlikht et al. / Engineering Fracture Mechanics 71 (2004) 19972013 Table 1 Critical CTOA for X80 grade pipe material Pipe diameter [m] 1.4 1.2 1.4 1.2 Wall thickness [mm] 18 16.3 26 13.8 CTOA [deg] 11.9 11.414.5 8.510.5 11.2 1.4 Experimental technique Two-specimen method 3-point bend test Full-scale burst test, crack ank opening measurement New specimen, direct CTOA measurement Ref. [12] [14] [30]

2005

Current work

growth is established and then maintained in this specimen over a considerable distance the CTOA has a tendency to rise again towards the end of the test, an eect associated with plastic hinge formation in the remaining ligament. From the data shown in Fig. 9 it appears that a steady state is achieved when the crack reaches the length of approximately two specimen thicknesses and continues for another six. The mean and standard deviations for stable state CTOA values obtained with both techniques are also shown in Fig. 9. The direct technique produces a CTOA value which is 1.5 higher than that produced by the indirect one on three pairs of grid lines. Higher scatter in indirect CTOA values is associated with the uncertainty in positioning the measuring lines on the grid lines due to the thickness of the latter. The results shown in Fig. 9 t quite well with those reported in the literature. The steady state in thin sheet 2024-T3 aluminium alloy [8] was established after the crack grew 12 thicknesses. A comparison between several reported CTOA values for typical X80 grade pipes is set out in Table 1. Except for the present work all results in this table were obtained with techniques in which the angle itself was not measured but rather deduced from other quantities measured in the experiments. The condence interval for the present work was calculated for 95% condence level. It is assumed in the approach adopted in [12] that the total energy to fracture of a notched specimen is proportional to its ligament length. By conducting the test on two specimens with dierent ligaments the slope of the proportionality line can be obtained. This slope is linked to CTOA by an equation proposed by the same group of authors earlier. A concept of a point of rotation in a three-point bend test was used in [14]. This concept leads to a similar approach in which CTOA is proportional to the slope of a straight line tted to the experimental data in the form of load line displacement versus logarithm of ligament length. In [30] CTOA was calculated from the displacement eld behind the crack tip reconstructed from the strain gauge records obtained during the full scale burst test. The uncertainty associated with the exact crack tip location is the major source of scatter in the CTOA value provided by this method. Even so, it provides a very illustrative point of comparison for the laboratory-derived values reported here.

4. Specimen development Extensive work on specimen design and CTOA measurement procedure was required before the results presented in Section 3 were achieved. Fig. 10 shows the original specimen design, with a thin ligament between two thicker loading arms. This design was intended to promote constraint in the test section. Specimens with three ligament thicknesses, 4, 8 and 10 mm, were machined, with three specimens of each thickness. The lengths of initial notches were 36, 80 and 96 mm accordingly. Each specimen had its initial notch extended about 2 mm by fatigue precracking prior to the tests according to Appendix A2 of ASTM E399-90 standard [31].

2006

A. Shterenlikht et al. / Engineering Fracture Mechanics 71 (2004) 19972013

Fig. 10. Initial specimen design. All dimensions are in millimeters.

4.1. Four millimetre thick specimens Four of the captured images obtained on one of the specimens are shown in Fig. 11. The opening of the fatigue grown part of the crack can be seen in Fig. 11a. The crack propagates in a slant mode from the very beginning (Fig. 11b). Fig. 11c shows the moment of change of inclination angle from +45 to )45. Finally Fig. 11d displays the behaviour of a crack when it deviates towards one of the loading arms. A photograph of the specimen after the test is shown in Fig. 12. The strong residual plastic deformation of the upper arm is clearly visible. The shape of the specimen suggests that the minor (horizontal) plastic strains were positive.

Fig. 11. Four of the captured frames of crack propagation in 4 mm thick specimen.

A. Shterenlikht et al. / Engineering Fracture Mechanics 71 (2004) 19972013

2007

Fig. 12. Photograph of the 4 mm thick specimen after the test.

4.2. Eight and ten millimetre thick specimens Unstable crack growth was observed during the test of a 10 mm thick specimen. The crack deviated into the loading arm immediately after the fatigue pre-crack part (Fig. 13). Such behaviour of thin and thick ligament specimens is explained in the next section in terms of T -stress. 4.3. T-stress discussion and the modied specimen The T -stress is the non-singular term in the expansion for the elastic crack tip eld. It acts parallel to the plane of the crack and depends on the shape and size of the cracked specimen [3234]. The evolution of T stress with crack growth has been calculated for dierent specimen geometries in [35]. From many cracked specimen geometries analysed in [35] the double cantilever beam (DCB) is the one with the greater similarity to the initial specimen design of this work.

Fig. 13. Photograph of the 10 mm thick specimen after the test.

2008

A. Shterenlikht et al. / Engineering Fracture Mechanics 71 (2004) 19972013

The results in [35, Fig. 8d] show that a positive (tensile) T -stress exists in this specimen during the test. The relative value of T -stress, T , (where r is the nominal stress acting across the crack section) grows slightly r from approximately 0.540.56 and then gradually falls to 0.3 as the crack progresses. The mechanical interactions that force the initial growth of T -stress can be associated with the observed crack path deviation towards one of the loading arms seen in Fig. 12. The moment when the crack touches the upper loading arm (Figs. 11d and 12) is associated with the crack tip passing through the peak value of T -stress. The signicant dierence in thickness between the test area and the arm for the 4 mm thick specimens (4 mm against 11 mm) prevents the crack from propagating into the arm at this point. After this point the T stress decreases, thus preventing the crack from deviating into the arm during the later stages of crack growth. A longer initial notch in a 10 mm thick specimen resulted in a specimen geometry where the crack starts from a point of maximum T -stress (compare Figs. 12 and 13). Consequently, the crack deviates in this 10 mm thick specimen as soon as it begins to grow from the initial fatigue crack. The dierence in thickness between the test area and the arm (10 mm against 11 mm) is not enough to prevent the crack from propagating into one of the loading arms. The only way to decrease the level of T -stress in this initial specimen design is to make it either shorter or wider [35, Fig. 8d]. The latter is more desirable since the length of the specimen is crucial in establishing the steady-state crack growth. But the width of the specimen can be increased only slightly (due to the curvature of the pipe from which the at specimen comes) and any such change is also associated with a decrease in at-plate thickness which is equally undesirable. Therefore the eective width of the specimens was increased by a factor of three with the use of loading plates. This modication brings the relative maximum T -stress down to 0.3 (or even to negative (compressive) values if the single edge cracked plate tension (SECT) model is used for the modied specimen geometry [35, Fig. 4b]). These considerations led to the modied specimen geometry, discussed in Section 2. However, this modication alone in 10 mm thick specimens did not prove suitable. Initial tests produced excessive plastic deformation of material close to the third loading hole (Fig. 14) and, in one instance shearing of material between the crack tip and the third hole (Fig. 14b). These observations led to the conclusion that 10 mm thick specimens are not suitable for this test and that an 8 mm working section thickness is the maximum safe value for this test that can be extracted from pipes of the order of 1.2 m diameter with 14 mm thickness. 4.4. CTOA results Fig. 15 shows all the CTOA data collected in the developmental 4, 8 and 10 mm specimens plotted against dimensionless crack length. Although the scatter is much greater than that in Fig. 9, it is possible to trace some features of CTOA evolution with crack growth. High CTOA values at the beginning of crack

Fig. 14. Photograph of the two of the modied 10 mm thick specimens after the test.

A. Shterenlikht et al. / Engineering Fracture Mechanics 71 (2004) 19972013


27.5 25 22.5 20 CTOA, deg 17.5 15 12.5 10 7.5 5 2.5 0 0 5 10 15 20 25 Crack length/specimen gauge thickness 30

2009

4 mm 8 mm 10 mm

Fig. 15. CTOACrack length results. The crack length is measured from the notch.

growth and towards the end of the test (only for 8 mm thick specimens) are clearly visible. There are anomalously high CTOA at about half the total crack path length for one of the 4 mm thick specimens. These are due to crack deviation towards the loading arm in this test (Figs. 11d and 12). The large scatter in the data is primarily due to the poor quality of the images captured from the video recording. (The quality of imaging was very much higher in the tests reported in Section 3.) According to the data of Fig. 15, the stable state CTOA value is in the range 812, which is smaller than the value 12.5 presented in the Section 3 (Fig. 9). This is possibly due to the fact that the plates from which the specimens were machined were cut from dierent circumferential positions in the pipe. Signicant variation in mechanical properties (i.e. in Charpy toughness) can occur in pipeline steels such as X80 with circumferential position [36]. The importance of scale length, x, (Fig. 7) was emphasized in Section 3. In these particular tests, this is even more important because of the stair-like appearance of the crack anks in some (but not all) of the tests (Fig. 16a). This was probably due to the inuence of the scored grid. Although the grid has apparently little, if any, inuence on the fracture propagation behaviour, such appearance of the crack prole results in additional inaccuracy in CTOA values. The crack growth as observed on the specimen surface is not constant. Several instances were observed when the crack jumps for several millimetres almost instantaneously. In some cases a small crack could be seen ahead of the main crack tip which then joined the main crack after the ligament between them broke. This behaviour made the determination of the crack tip problematic. One illustration of this phenomenon can be seen from Fig. 16b. The main crack is shown growing from left to right with the crack tip marked (A) and another small crack marked (B), just ahead of it. The distance between the main crack and the small one is 1 mm. A short time after this image was captured the remaining ligament fractured and the small crack became a part of the main one. A sensible compromise is for the optimal value of x in Eq. (2) to be at least two times bigger than the average stair step size, which usually was 13 mm. Therefore the values x 36 mm were used depending on the crack tip appearance in each frame. The CTOA values shown in Fig. 15 were obtained with both direct and indirect techniques. The indirect technique is potentially more powerful and can give reliable results even on jagged crack anks or poor quality images. Fig. 17 shows one such image registered on a 10 mm thick specimen. The CTOA

2010

A. Shterenlikht et al. / Engineering Fracture Mechanics 71 (2004) 19972013

Fig. 16. Illustration of the complications in CTOA measurement. (a) Photograph of a fragment of a fracture surface with stair-like crack edges. Beachmarks on the fracture surface are also clearly visible. (b) One of the captured video images showing the tip of the main crack (A) and another small crack just ahead of it (B). They are separated by one grid step (1 mm).

measurement on the crack anks in this image is very dicult because the exact location of the crack tip is very uncertain and the image resolution is poor. A direct CTOA value can be obtained if a point A is tted somewhere around the perceived crack tip and two points B and C are located at distance x behind on the crack anks. For this image, this direct technique gives 1620, depending on the exact location of points A, B and C. For the indirect CTOA measurement, two vertical lines DJ and EK are drawn at some distance behind and ahead of the perceived crack tip. The exact location of the crack tip is not now important. The point at which these lines cross the grid lines are marked and the CTOA can be obtained from trapezoids DFGE (13.4), DHIE (11.0) and DJKE (9.3) for the rst, second and third pair of grid lines accordingly. Not only are the indirect CTOA measurements more precise than the direct one in this particular case, but also a gradual decrease in CTOA with increasing order of the grid line used provides additional check for the reliability of this measure.

Fig. 17. Comparison between direct and indirect CTOA measuring techniques in the presence of highly jagged crack anks. For explanation see text.

A. Shterenlikht et al. / Engineering Fracture Mechanics 71 (2004) 19972013

2011

4.5. Conclusions lead to the nal test conguration As a result of this preliminary test work, the following decisions were taken. The thickness of the test area was chosen to be 8 mm as a compromise between the desire to make it thicker and therefore closer to the pipe wall thickness and the need to make it small enough to ensure stable crack growth. The friction between the specimen and the loading plates should be as high as possible to bring the loading conditions on the edges of the specimen closer to uniform ones. This will decrease the inuence of the holes in the loading arms on CTOA i.e. decrease the scatter and periodic uctuations (Fig. 9). The side surface of the specimen has to be free from scratches since they might interfere with the crack appearance on the surface and therefore aect the CTOA values. Thus it was decided to paint the specimen side surface and to score the grid on the paint. The painting of the surface has a further positive eect in reducing light reection from the specimen surface. The indirect CTOA measuring technique is at least as reliable as the direct one and in certain situations it should be given preference.

5. Conclusions A new DCB-like specimen induces large amounts of stable ductile crack growth in a pipe material. The specimen may be cut directly from pipe without any subsequent attening and is oriented on direct optical CTOA measurement. A coat of dark matt blue paint sprayed onto the surface can be used to suppress reection. A rectangular grid scored in the surface paint eases the measurement of CTOA and increases its accuracy comparing with alternative methods. The maximum gauge thickness producing stable crack growth was 8 mm for the X80 pipe material used in this work. CTOA was measured both on a crack prole and on the grid lines and the resulting values are 1.5 less for the latter than for the former. Also the scatter is slightly higher for CTOA data obtained via the indirect technique. The major sources of uncertainty are the exact crack tip location in the direct technique and the thickness of grid lines in the indirect one. Special care must be taken in ensuring the maximum grip between the loading plates and the specimen since amounts of even small sliding can cause unstable crack growth or result in large variance in CTOA. A key factor in obtaining accurate CTOA measurements is good optical equipment.

Acknowledgement The work was sponsored by Advantica Technologies Ltd.

References
[1] Gillemot F, Havas I, Rittinger J, Fehrvry A. Experiences in the comparison of large and small fracture mechanics specimens. In: e a Kussmaul K, editor. Fracture mechanics verication by large-scale testing, EGF/ESIS8. London: Mechanical Engineering Publications; 1991. p. 319. [2] Anderson TL, McHenry HI, Dawes MG. Elasticplastic fracture toughness tests with single-edge notched bend specimens. In: Wessel ET, Loss FJ, editors. Elasticplastic fracture test methods: the users experience. ASTM STP, vol. 856. American Society for Testing and Materials; 1985. p. 21029.

2012

A. Shterenlikht et al. / Engineering Fracture Mechanics 71 (2004) 19972013

[3] Bilby BA, Howard IC, Li ZH. Prediction of the rst spinning cylinder test using ductile damage theory. Fatigue Fract Engng Mater Struct 1993;16:120. [4] Bilby BA, Howard IC, Othman AM, Lidbury DPG, Sherry AH. Prediction of spinning cylinder tests 2 and 3 using continuum damage mechanics. In: Mehta HS, editor. Fracture mechanics applications, PVP-vol. 287/MD-vol. 47. The American Society of Mechanical Engineeers; 1994. p. 310. [5] Howard IC, Li ZH, Sheikh MA, Lidbury DPG, Sherry AH. The simulation of the fourth spinning cylinder test using damage mechanics. In: Yoon K, editor. ASME PVP Conference, Montreal, 2126 July. Fatigue and Fracture, vol. 2. The American Society of Mechanical Engineeers; 1996. p. 25765. [6] Burstow MC, Howard IC, Pugh AC, Yates JR. Prediction of the toughness of a wrought stainless steel using ductile damage mechanics properties obtained from small samples. In: Lidbury D, Kirk M, Mohan R, editors. Applications of fracture mechanics in failure assessment, PVP-vol. 412. The American Society of Mechanical Engineeers; 2000. p. 5361. [7] Burstow MC, Howard IC. Predicting the eects of crack tip constraint on material resistance curves using ductile damage theory. Fatigue Fract Engng Mater Struct 1996;19(4):46174. [8] Dawicke DS, Sutton MA. CTOA and crack-tunneling measurements in thin sheet 2024-T3 aluminum alloy. Exp Mech 1994;34(4):35768. [9] Dawicke DS, Newman Jr JC, Bigelow CA. Three-dimensional CTOA and constraint eects during stable tearing in a thin-sheet material. In: Reuter WG, Underwood JH, Newman Jr JC, editors. Fracture mechanics. ASTM STP 1256, vol. 26. American Society for Testing and Materials; 1995. p. 22342. [10] Dawicke DS, Piascik RS, Newman Jr JC. Analysis of stable tearing in a thick aluminum alloy. In: Underwood JH, Macdonald BD, Mitchell MR, editors. Fatigue and fracture mechanics. ASTM STP 1321, vol. 28. American Society for Testing and Materials; 1997. p. 30924. [11] Jones R, Rothwell AB. Alternatives to charpy testing for specifying pipe toughness. In: Cutler J, Rothwell AB, editors. Fracture control in gas pipelines. WTIA/APIA/CRC-MWJ International Seminar, WTIA/APIA/CRC-MWJ, 1997. p. 5-121. [12] ODonoghue PE, Kanninen MF, Leung CP, Demofonti G, Venzi S. The development and validation of a dynamic fracture propagation model for gas transmission pipelines. Int J Pressure Vessels Piping 1997;70(1):1125. [13] Pussegoda N, Malik L, Dinovitzer A, Graville BA, Rothwell AB. An interim approach to determine dynamic ductile fracture resistance of modern high toughness pipeline steels. In: Ellwood JR, editor. Proceedings of the 2000 International Pipeline Conference, vol. 1. New York: The American Society of Mechanical Engineeers; 2000. p. 23945. [14] Pussegoda N, Verbit S, Dinovitzer A, Tyson W, Glover A, Collins L, et al. Review of CTOA as a measure of ductile fracture toughness. In: Ellwood JR, editor. Proceedings of the 2000 International Pipeline Conference, vol. 1. New York: The American Society of Mechanical Engineeers; 2000. p. 24754. [15] Mannucci G, Harris D. Fracture properties of API X 100 gas pipeline steels, Tech. Rep. EUR 20330 EN, European Commission, Directorate-General for Research, Brussles, Belgium, 2002. [16] Wilkowski G, Wang Y-Y, Rudland D. Recent eorts on characterizing propagating ductile fracture resistance of linepipe steels. In: Denys R, editor. Pipeline technology, vol. 1. Netherlands: Elsevier; 2000. p. 35986. [17] Andrews RM, Howard IC, Shterenlikht A, Yates JR. The eective resistance of pipeline steels to running ductile fractures; Modelling of laboratory test data. In: Neimitz A, Rokach IV, Kocada D, Goo K, editors. ECF14, Fracture mechanics beyond n s 2000. Sheeld, UK: EMAS Publications; 2002. p. 6572. [18] Andrews RM, Chterenlikht A, Howard IC, Yates JR. Measurement and modelling of the crack tip opening angle in a pipeline steel. In: IPC2002, Proceedings of the 2002 International Pipeline Conference. New York: The American Society of Mechanical Engineeers; 2002. p. 27062.18. [19] Abou-Sayed IS, Marshall CW, Kanninen MF. An assessment of the fracture toughness associated with at and slant crack growth in A533B steel. In: Francois D, editor. Advances in fracture research, ICF/SF2M. Oxford: Pergamon Press; 1981. p. 22734. [20] Miglin MT, Hirth JP, Roseneld AR. Estimation of transverse-shear fracture toughness for an HSLA steel. Int J Fract 1983;22(2):R657. [21] Schroth JG, Hirth JP, Hoagland RG, Roseneld AR. Combined mode Imode III fracture of a high strength low-alloy steel. Metall Trans A 1987;18A:106172. [22] Kumar AM, Hirth JP. Mixed mode I/III fracture testing. Scr Metall Mater 1991;25:98590. [23] Gordon JA, Hirth JP, Kumar AM, Moody Jr NE. Eects of hydrogen on the nixed mode I/III toughness of a high-purity rotor steel. Metall Trans A 1992;23A:10139. [24] Kumar AM, Moody NR, Hirth JP, Gordon JA. Eects of hydrogen on the mixed mode I/III toughness of a high-purity rotor steel. Metall Trans A 1993;24A:14501. [25] Kurita Y, Akiyama T, Kitao K. Evaluation of shear crack arrest capability of control rolled line pipe steel. In: Sih GC, Mirabile M, editors. Analytical and experimental fracture mechanics. Kluwer Academic; 1981. p. 53952. [26] Rivalin F, Iung T, Di Fant M, Pineau A. Dynamic ductile tearing in high strength pipeline steels. In: Proceedings of the ASME International Pipeline Conference. The American Society of Mechanical Engineeers; 1996. p. 17782. [27] Rivalin F, Pineau A, Di Fant M, Besson J. Ductile tearing of pipeline-steel wide plates. I. Dynamic and quasi-static experiments. Engng Fract Mech 2001;68(3):32945.

A. Shterenlikht et al. / Engineering Fracture Mechanics 71 (2004) 19972013

2013

[28] Kylander OS, Kylander K. GIMP: The ocial handbook. USA: The Coriolis Group; 1999. [29] Gao X, Shih CF. A parametric study of mixed mode I/III ductile fracture in tough materials under small scale yielding. Engng Fract Mech 1998;60(4):40720. [30] Berardo G, Salvini P, Mannucci G, Demofonti G. On longitudinal propagation of a ductile fracture in a buried gas pipeline: numerical and experimental analysis. In: Ellwood JR, editor. Proceedings of the 2000 International Pipeline Conference, vol. 1. New York: The American Society of Mechanical Engineeers; 2000. p. 28794. [31] ASTM E 399, Standard Test Method for Plane-Strain Fracture Toughness of Metallic Materials, American Society for Testing and Materials, Philadelphia, 1990. [32] Bilby BA, Cardew GE, Goldthorpe MR, Howard IC. A nite element investigation of specimen geometry on elds of stress and strain at the tip of stationary cracks. In: Size eects in fracture, Institution of Mechanical Engineers. Mechanical Engineering Publications; 1986. p. 3746. [33] Bilby BA, Goldthorpe MR, Howard IC, Li ZH. Size and constraint eects in ductile fracture. In: Dyson BF, Hayhurst DR, editors. Materials and engineering design, The next ten years. London: Institute of Metals; 1989. p. 20512. [34] Beteg n C, Hancock JW. Two-parameter characterization of elasticplastic crack-tip elds. J Appl Mech 1991;58:10413. o [35] Sherry AH, France CC, Goldthorpe MR. Compendium of T -stress solutions for two and three dimensional crack geometries. Fatigue Fract Engng Mater Struct 1995;18:14155. [36] Andrews RM. Private communication. Advantica Technologies Ltd; 2002.

Вам также может понравиться