You are on page 1of 10

Proceedings of the 2012 9th International Pipeline Conference IPC2012 September 24-28, 2012, Calgary, Alberta, Canada

IPC2012-90637

EVALUATION ON DEPENDENCE OF DUCTILE CRACK PROPAGATION RESISTANCE ON CRACK VELOCITY


Shuji Aihara The University of Tokyo Bunkyo-ku, Tokyo, Japan Yasuhito Imai The University of Tokyo presently, Tokyo Gas Co., Ltd. Taishi Fujishiro Nippon Steel Corp. Kimitsu, Chiba, Japan Kazuki Shibanuma The University of Tokyo Bunkyo-ku, Tokyo, Japan Takuya Hara Nippon Steel Corp. Kimitsu, Chiba, Japan

ABSTRACT Plastic strains were measured near a propagating ductile crack in drop-weight tear tests (DWTT). Plastic work evaluated from the plastic strains agreed with crack propagation energy evaluated from dynamic load versus displacement curve. Furthermore, plastic strains were measured near a propagating ductile crack in a full-scale burst tested pipes. Plastic work of the burst pipe was found to be much larger than that of the DWTT. Values of the plastic work of the DWTT and the burst pipe were plotted against crack velocity to construct crack resistance curve. Reason to the increased crack resistance with crack velocity was understood as a dependence of plastic strain distribution on crack velocity by elasto-plastic dymanic finite element analysis. INTRODUCTION Natural gas is one of the most promising primary energies due to its lower green house gas emission and less uneven distribution in the world than others. Natural gas transmission pipelines will be expected to increase steeply with increased demand for natural gas, in the near future. Needs for more efficient and longer transmission of natural gas prompt higher pressure operations. To realize this end, high strength linepipes up to X100 and X120 have been developed. One of the most crucial issues in securing the integrity of those high-pressure gas pipelines is the prevention of unstable ductile fracture. Procedures for predicting the unstable crack propagation and arrest behaviors have been developed and applied to pipeline design, e.g. [1][2]. Most widely used procedure is the Battelle two-curve, BTC, approach [1]. The BTC is simple; gas decompression curve and crack resistance curve are compared and it is judged that a crack does not propagate if these two

curves do not intersect each other. However, recent full-scale burst tests indicated that the BTC does not necessarily give correct answers, especially for high-pressure and high-strength pipelines [3]. The HLP approach which was developed by the High-pressure Line Pipe research committee Japan is one of the alternatives [2]. The original HLP procedure is based on fullscale burst tests of X70 linepipes and it is said that it gives better predictions at high grade linepipes. However, it still needs modifications for better accuracy [4]. Both the BTC and HLP approaches are based on full-scale burst tests data and necessarily depend on the data set. Therefore, these approaches might lower their accuracy in the prediction for different grade, pressure, pipe diameter etc. One of the authors developed a numerical model which predicts unstable ductile crack propagation and arrest behaviors in gas pipelines [5]. Their model is not based on full-scale burst test data but on physical assumptions. So, it can also be applied to such unconventional pipelines like hydrogen gas pipelines [6]. The model, as well as the conventional approaches, needs crack resistance curve. The model uses crack resistance value measured by drop-weigh tear test, DWTT, and must assume its dependence on crack velocity. Accurate assumption of the crack resistance curve is definitely important for better accuracy. At this moment, the model assumes the crack resistance curve based on a few full-scale burst test data. A more universal procedure to determine the crack resistance curve is necessary. From the above standpoint, the authors conducted DWTT tests using X65 linepipe steels and conducted detailed measurement of plastic deformation near a propagating crack and discussed the relationship between the plastic work and DWTT absorbed energy. In addition, plastic deformations near a propagating crack in full-scale burst tests were measured and

Copyright 2012 by ASME

compared with those of the DWTT. From these experiments, a method to construct a crack resistance curve is proposed. To investigate fundamental aspect of dynamic ductile crack propagation, elasto-plastic dynamic finite-element analysis was conducted and compared with the test results. DROP-WEIGHT TEAR TESTS Tested Steel The steels tested in the present study are API 5UL-X65 linepipes, whose chemical composition is shown in Table 1. Thermo-mechanical control process, TMCP, and thermo mechanical rolling process, TMR, were applied in the rolling process and 13.5 mm thick plates were produced. UOE pipes with outer diameter of 559 mm were produced from the steels. Table 2 shows their mechanical properties. Table 3 shows Charpy impact and DWTT results conducted by the mill. Table 1 Chemical composition of the steel tested. C Si Mn P S Nb Ceq 0.04 0.33 1.41 0.008 < 0.001 0.03 0.34 Table 2 Mechanical properties of the pipes tested. YS TS U-El El YR [MPa] [MPa] [%] [%] 571 665 4.5 32 86 506 583 10 38 87 Charpy impact and DWTT test results at 0 C. Charpy impact DWTT Absorbed Absorbed Shear % Shear % Energy [J] Energy [J] 492 100 10,421 100 467 100 9,131 100

Results All the specimens exhibited 100 % shear fracture, same for the full-scale burst test explained later. Figure 1 shows the specimen surfaces of the specimen, #5 (20 C) and #9 (0 C), after the test. The deformed pattern clearly indicates intense plastic deformation near the crack path. Figure 2 shows dynamic load versus displacement curves. No large differences were observed between the specimens and testing temperatures. Absorbed energy was evaluated by integrating the load versus displacement curve, result of which is shown in Table 4. Note that the value in terms of the unit [J] is the integrated value of the load versus displacement curve and that of [MPa m] is the integrated value divided by the uncracked ligament area. and are the energy up to and after the maximum load and regarded as initiation and propagation energy, respectively. is the total absorbed energy. Value of ranged 6.1 to 6.4 MPa m and no considerable influence of testing temperature was observed. Figure 3 shows ductile crack length measured from the images taken by the high-speed camera. In the authors experiments different from the present study, the front of a shear crack was identified from the observation of arrow-head markings or a series of micro-separations, which showed almost straight crack front, although the front of a flat ductile

TMCP TMR Table 3

TMCP TMR

Experimental Standard DWTT specimen blanks were machined in the transverse direction, with the exception that specimen thickness was reduced to 12 mm and a notch with a width of 0.2 mm was machined. The notch depth was 5 mm. The machined notch was applied so that ductile fracture was easily initiated. Moreover, the specimen surface was electrolytically etched to print circle patterns, diameter of which was 5 mm. This was to measure plastic deformation after the tests. Four DWT tests were conducted by a drop weight tower with a weight of 12.5 kN and drop height of 4 m. Impact velocity was 8.9 ms. The tests were conducted at 20 C for the specimen #5 and #6, and 0 C for #9 and #10. The testing machine was equipped with a high speed camera for measuring crack propagation length dynamically, as well as dynamic load and displacement measurement by a load cell and a laser displacement measuring system. After the tests, the specimens were subjected to measurement of diameters of the deformed circle patterns and strains were evaluated.

(a)

#5 (20 C)

(b) Fig. 1

#9 (0 C)

Deformation pattern after the test.

Copyright 2012 by ASME

250 200 Load [kN] 150 100 50 0 0 20 40 60 Displacement [mm] 80 Crack length [mm] #5 (20C) #6 (20C) #9 (0C) #10 (0C)

80 60 40 20 0 0 20 40 60 Displacement [mm] 80 #5 (20C) #6 (20C) #9 (0C) #10 (0C)

Fig. 2 Load versus displacement curves.

Fig. 3

Ductile crack length measured by high-speed camera.


12.0

Test Piece #5 #6 #9 #10

Table 4 Absorbed energy, TMCP Initiation Propagation Total [J]/[MPa m] [J]/[MPa m] [J]/[MPa m] 2,618 5,455 8,073 3.10 6.38 9.48 2,606 5,409 8,015 3.05 6.33 9.38 2,710 5,224 7,925 3.17 6.11 9.28 2,719 5,294 8,013 3.18 6.20 9.38

J-resistance [MPa m]

10.0 8.0 6.0 4.0 2.0 0.0 0 5 10 15 20 25 Crack length [mm] 30 35 #5 (20C) #6 (20C) #9 (0C) #10 (0C)

Fig. 4 crack initiated from the precrack tunneled [7]. The crack was initiated at load point displacement of approximately 13 mm and the crack length increased almost linearly with displacement. From the measured load versus displacement and crack length versus displacement curves, J-integral resistance curve was evaluated by the nest equation [8], 2 = { (0 + )} (0 + ) (1) =

J-resistance curves.

0 0 0 = 0 1 = 1 (1 + )(1 + )

(2)

where, is crack length, is load, is load point displacement, is specimen thickness, is specimen width ( = 75 mm ), 0 is initial crack length ( = 5 mm ), is ductile crack length. It should be noted that the equation should be applied to a deeply cracked bend specimen and therefore the evaluated J-resistance curve should be regarded as approximation. Figure 4 shows the J-resistance curves evaluated by this way. Value of at crack initiation was 4 MPa m, approximately. It should be noted that this is only for reference due to the reason mentioned above. The J resistance value increased with increasing crack length. The resistance value ranged 6 to 10 MPa, approximately, at ductile crack length of 10 to 30 mm. Strain distributions were evaluated from the deformed circle patterns on the specimen surface by the following equations,

where, 0 is circle diameter before the test (= 5 mm), and are diameters in the crack propagation and transverse direction, respectively. Thickness direction strain, , was evaluated by the last equation, which is derived from the incompressibility of plastic deformation. Although shear strain could be evaluated from angular distortion of the grid lines, it was neglected due to much smaller values than the normal strains. Equivalent strain was evaluated from the strain components by, 2 2 2 2 2 = ( + + + ) 3

(3)

Figures 5 (a) through (e) show strain distributions at different crack path locations in the specimen #5. Note that the horizontal axis is the position transverse to the crack direction, . Plastic deformation extended about 20 mm from the crack position. Value of was much lower than that of , and

Copyright 2012 by ASME

(a)

#5 (20 C), = 10 mm.

(b)

#5 (20 C), = 20 mm

(c) #5 (20 C), = 30 mm

(d) #5 (20 C), = 40 mm.

(e)

#5 (20 C), = 50 mm

(f)

#6 (20 C), = 20 mm.

(g)

#9 (20 C), = 20 mm

(h) #10 (20 C), = 20 mm

Fig. 5 Strain distributions in the DWTT (: distance from notch tip). was negative. Value of is predominantly determined by . Figures 6 (f) through (h) show strain distributions at 20 mm from the notch-tip in the specimens #6, #9 and #10, respectively. The strain distributions in these specimens are almost the same with that of the specimen #5. For evaluating strain energy density associated with the plastic strains shown above, stress-strain curves were obtained from tensile tests at subzero temperatures [9]. Tensile specimens according to JIS Z 2201 type-5 were machined in the transverse direction. True stress-strain curves were obtained from the test at room temperature (15 C) and fitted to the Swift formula, Eq. (4) where, 0 = 0.005. Table 5 shows the fitted parameters. Y () = Y0 ( + 0 ) 0 (4) Table 5 Strain hardening parameters. Strain Hardening 0.2 % Proof Stress Exponent Y0 [MPa] 585 0.050 488 0.114

TMCP TMR Table 6

Strain rate and temperature dependence parameters. 0 [MPa] TMCP 479 1.52 x 103 TMR 391 1.76 x 103

taken into account. It is well known that the strain rate and temperature effect on yield strength is equivalent because of the thermal activation process of dislocation motion. Hence, the following equation can be justified [9], Y = 0 exp ( ) (5)

Because strain rate associated with unstable ductile fracture is quite high, elevation of yield strength should be

Copyright 2012 by ASME

8.0 Plastic work [MPa m]

6.0

4.0

2.0

0.0 0 10 20 30 40 50 60

Fig. 6

Distributions of strain energy density in the transverse direction, #5 (20 C).

Distance from notch tip [m]

Fig. 7

Distribution of plastic work in the transverse direction, #5 (20 C).


10.0 Plastic work [MPa m]

where, R is strain rate-temperature parameter, as = ln ( 10 )


8

Propagation energy Ep
8.0 6.0 4.0 2.0 0.0 0.0 2.0 4.0 6.0 8.0 10.0 Absorbed energy [MPa m]

Total energy Et

(6)

where, [K] is temperature and [1s] is strain rate. Table 6 shows the values of 0 and k obtained from the tensile tests. It is quite difficult to determine the strain rate for unstable ductile fracture. But, if we assume that crack velocity is 10 ms , which roughly corresponds to that of the present DWTT, and the strain achieved near the crack-tip is 0.4 for DWTT, then = 2 102 [1s] as an order of magnitude estimate. We employ this value in the present study and the yield strength was modified according to Eq. (5). Then, the strain energy density can be estimated by,

Fig. 8

Comparison between plastic work and absorbed energy.

( ) =

(7) that the tangent of a J-resistance curve is related to the plastic work associated with crack propagation, then nearly constant tangent at 10mm to 30mm crack propagation might be consistent with nearly constant plastic work at this crack length, Fig.7. Assuming that steady state is maintained at 20mm crack propagation in this specimen as well configuration, at 20 mm was calculated for the other specimens and compared with the absorbed energy evaluated from the load versus displacement curve, as was shown in Table 4. Figure 8 shows the result for the four specimens. We can see good agreement between and Ep. It should be noted that Et estimate the plastic work excessively because it contains the crack initiation energy. It should also be noted that the J-resistance value, about 9 MPam at 20mm crack extension, roughly corresponds to and . The discrepancy between , and J-resistance value might be due to inappropriate specimen configuration for J evaluation; the J-resistance curve should be evaluated by using a deeply cracked specimen, as mentioned previously.

Figure 6 shows distributions of the strain energy density in the transverse direction in the specimen #5. It has maximum at crack position and extends as far as about 20 mm in the transverse direction. Moreover, it changes with distance from the notch-tip. Plastic work or energy dissipation associated with crack propagation can be evaluated by integrating the strain energy density in the transverse direction,

(8)

Result for the specimen #5 is shown in Figure 7. Although the value of changes with distance from the notch-tip, it is almost constant at 10 to 30 mm and becomes lower at longer crack propagation. Because the J-resistance curves were obtained with the specimens with quite shallow notch, those of Fig.4 should be regarded as a reference. However, if we assume

Copyright 2012 by ASME

Table 7 Burst test conditions. Grade of Tested Pipe API 5L-X65 UOE Outer Diameter 559mm Wall Thickness 13.5mm Pressurizing Medium 99.9% Hydrogen Line Pressure 16.0MPa Hoop Stress 72% SMYS Initial Crack Length 700mm (aimed)

TMR
E4P E3P E2P E1P

TMCP Side P
W1P W2P W3P W4P

100 mm

Explosive charge

E4N E3N E2N E1N

W1N W2N W3N W4N

West

Initial crack

East

Side N

(TMR) 44 m

(TMCP)

Fig. 11

Alignment of the circle and grid patterns.

Fig. 9 Layout of the full-scale burst test [6].

y x

(a) (a) Circle pattern

Circle and grid pattern


C (xC, yC) D (x'D, C (x'C, y'C)

D (xD, yD)

y x A (xA, yA)
B (xB, yB) B (x'B, y'B) A (x'A,

(b) Grid pattern Fig. 10 Circle and grid patterns drawn on the pipe surface

(b)

Definition of displacement

Fig. 12 Circle and grid patterns for plastic strain measurement.

FULL-SCALE BURST TESTS Experimental Conditions Plastic deformation associated with unstable crack propagation in a full-scale burst test was measured. Detail of the test is described elsewhere [6]. Tested pipes were just the same as those of the present study, TMCP and TMR. Table 7 summarizes the testing conditions and Figure 9 shows the test layout. Strain Measurement Procedure For the purpose of measuring the plastic strains near the propagating crack, circle and grid patterns were printed and drawn on the pipe surface. Figures 10 (a) and (b) show the both patterns, respectively. Circles with a diameter of 5 mm were printed by a rubber stamp. On the other hand, the grid lines with interval of 10 mm were printed and drawn by scribing

the pipe surface with a sharp scriber. The circle patterns and the grid patterns were prepared between the timing wires for crack velocity measurement, see Figure 11. Deformations of the circle and grid patterns were measured after the burst test. Normal plastic strain in the axial direction and in-plane plastic shear strain was evaluated by the next equations, see Figures 12 (a) and (b). =
(B B ) (A A ) B A

(9)

1 ( ) (A A ) (B B ) (A A ) = { } 2 A B A

(10)

Because the pipe received circumferential bending deformation, it is unreasonable to evaluate the normal plastic strain in the transverse direction, , from the deformation at the pipe

Copyright 2012 by ASME

E4P E3P

E2P E1P

W1P W2P W3P W4P

E4N W3N W4N

W2N

E3N

E2N

E1N

W1N

Fig. 13

Positions of the plastic strain measurement.

(a)

Circle pattern

(b) Grid pattern Fig. 14 Deformed patterns after the test.

surface. Hence, thickness contraction was measured by a micrometer and plastic strain in the thickness direction, , was evaluated. And then, was evaluated by assuming the incompressibility of plastic deformation, = 1 1 (1 + )(1 + ) (11)

Results Detail of the test result is described in [6]. Note that the whole crack propagation process finished before the decompression wave reflected at the pipe ends arrived at the

test section. So, the test can be regarded as valid. Fracture surface was 100 % shear. Because the crack length was not long, less than 1m, and the deformed patterns were not necessarily clear, measuring points were quite limited. Figure 13 shows positions for the measurement. Figures 14 (a) and (b) show deformed patterns near the propagating crack. Points W1P for TMCP and E1P, E2N/E2P (both sides at the same crack location) for TMR were subjected to the strain measurement. The circle pattern at W1P and E1P, and the grid pattern at E2N/E2P were employed. Figures 15 (a) through (c) show the evaluated strain distributions at these three points. Note that the strains were evaluated in the both sides of the crack at E2N/E2P (TMR). Due to the limitation of the micrometer, the measurement was not further than 120 to 160 mm. However, the Figures indicate quite wide spread of the plastic deformation in the transverse direction, which was much wider than that of the DWTT. In the full-scale burst test, was larger than , which was different from the strains in the DWTT. It is also noted that the shear strain was smaller than the other strain components. Interesting point is that maximum strain near the crack location, 0.4 to 0.6 , is comparable to that of the DWTT, 0.3 to 0.4, see Figure 5. Figure 16 shows distributions of strain energy density which was evaluated from the equivalent strain distribution and stress-strain curve, by the same manner as in Figure 6. Again, strain rate, = 2 102 [1s] , was assumed. At E2N/E2P (TMR), the distribution of the strain energy density was almost symmetrical on the both sides. Also, no considerable change with the location and the steels was observed. Value of was evaluated by integrating the strain energy density distributions in Figure 16. Note that the integrated value on the either side of the crack was doubled at W1P (TMCP) and E1P (TMR) assuming the symmetry. Because the plastic deformation extended beyond the measurement range, evaluated by this way must be a lower bound value. However, this value might be close to actual value because the plastic deformation beyond the measurement range is not so large. On the other hand, crack velocity was evaluated from the timing wire records [6] and those at the measuring locations were estimated. Figure 17 shows versus crack velocity. The Figure also includes the data points of derived from the DWTT. The value of for the full-scale burst test is much larger than that of the DWTT. In addition to the present analysis, the same was made for the full-scale burst tests and DWT tests conducted by one of the authors [10], using X65 ERW pipes. Here, the DWTT energy is not the value of but plateau value of the J-resistance curve. Figure 18 shows normalized by that of the DWTT versus crack velocity. For the tested three X65 steels, a large difference was not observed and an average curve can be drawn. The curve can be expressed as, = 0 [1 + 38.8 ( 3 ) ] 250 (12)

Copyright 2012 by ASME

0.6 0.4 0.2 Strain 0.0 -0.2 -0.4 -0.6 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 Distance from crack [m] Strain energy density [MPa]

500 400 300 200 100 0 -0.20 TMCP W1P TMR E1P TMR E2N & E2P

-0.10 0.00 0.10 Distance from crack [m]

0.20

(a)
0.6

W1P (TMCP)

Fig. 16
80 Absorbed energy [MPa m] 70 60 50 40 30 20 10 0

Distributions of strain energy density in the burst test.

0.4 0.2 Strain 0.0 -0.2 -0.4 -0.6 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 Distance from crack [m]

TMCP TMR

50

100 150 200 Crack velocity [m/s]

250

(b) E1P (TMR)


0.6 Normalized bsorbed energy 0.4 0.2 Strain 0.0 -0.2 -0.4 -0.6 -0.20 -0.10 0.00 0.10 0.20 20

Fig. 17

Plastic work versus crack velocity.


250
3

= 0 1 + 38.8 15 10

TMCP 5 0 0 50 100 150 200 Crack velocity [m/s] 250 TMR ERW

Distance from crack [m]

(c) E2N/E2P (TMR) Fig. 15 Evaluated strain distributions in the burst test.

Fig. 18

Plastic work normalized by that of DWTT versus crack velocity.

where, [ms] is crack velocity and 0 is crack resistance at zero crack velocity and can be approximated by of the DWTT. The value of might also be approximated by , as shown in Fig.8. NUMERICAL ANALYSIS AND DISCUSSIONS Model Description A dynamic elasto-plastic finite element analysis was carried out in order to investigate the influence of crack

velocity on the plastic work associated with crack propagation. The analysis was performed by using ABAQUS [11]. In the present study, constant velocity crack propagation in a plate was modeled under plane stress condition. Second order interpolation was used as the finite element approximation and infinitesimal deformation was assumed. The crack propagation was modeled by releasing reaction force of the crack tip node. A quarter-symmetry finite element model was used. Youngs module was 206 GPa, Poissons ratio was 0.3 and the density was 7,800 kgm3 . Yield stress was 600 MPa. Strain rate dependence of the yield stress was not taken into

Copyright 2012 by ASME

sy
Strain energy density [MPa]

150

Propagating crack

Plastic work evaluation line

120 90 60 30 0

V = 0 m/s, e f = 0.152 V = 500 m/s, e f = 0.154

sx

A a a = 10 m W = 137.6 m (a) Entire model (b) Detail of A Da = 50 mm

0.0

0.2

0.4 0.6 0.8 Distance from crack [m]

1.0

Fig. 19 Example of mesh division and boundary conditions in finite element model for = 200 ms, = 2,354 MPam.
Equivalent plastic strain: eeq 0.14 0.12 0.10 0.08 0.06 0.04 0.02 0.00

Fig. 21

Numerical results of distributions of strain energy density at 0.15.

fracture strain . Referring to the singular strain field of quasistatic crack propagation [12], the strain ahead of the crack tip was fitted by = 1 log + 2 (13)

50mm

50mm

(a) V = 100 m/s

(b) V = 500 m/s

Fig. 20

Numerical results of equivalent plastic strain field near crack tip at = 2,354 MPam.

consideration. Elastic linearly strain-hardening was assumed and a tangential stiffness modulus was 2,120 MPa. Assuming the pipe hoop stress equal to 0.7 times the yield stress and the axial stress of 0.5 , applied stress in the model was set as = 0.5 = 210 MPa and = = 420 MPa. Crack velocity was = 0 ~500 ms . In the case of = 0 ms, quasi static analysis was performed. The plastic work at a crack length was evaluated based on Eq. (7) on the line vertical to of the crack behind 50 mm from the crack tip. The entire finite element model size depended on and . In the case of = 0 ms, we determined = 10. In the cases of 0 ms, was determined so that the reflection of elastic wave at the model boundary does not occur. An example of mesh division is shown in Figure 21, where = 200 ms and = 10 m . Minimum mesh size was 5 mm. Results and Discussions Figure 22 shows an example of the calculated equivalent plastic strain filed. It is found that higher crack velocity gives smaller plastic deformation at the same stress intensity factor. Assume that a crack continues to propagate if tensile strain at a characteristic distance , from the crack-tip exceeds

where is distance from the crack tip. 1 and 2 are fitting parameters. at = was determined from the fitted curve where was assumed equal to 1 mm. Distributions of strain energy density at about the same condition of 0.15 are shown in Figure 23. It is clear that higher crack velocity makes larger plastic work if the same fracture strain is assumed. The elasto-plastic dynamic finite element analysis has shown that strain accumulation at a crack-tip decreases with increasing crack velocity. This phenomenon can be explained by a competition between plastic wave propagation and crack propagation, i.e., plastic deformation becomes more difficult with increasing crack velocity because the time period for plastic deformation to spread with limited plastic shear wave velocity becomes shorter at higher crack velocity [13]. Figures 22 and 23 clearly indicated this mechanism. Ductile fracture is strain-controlled type of fracture. As mentioned above, we can assume that a ductile crack continues to propagate under a condition that local strain at a crack-tip reaches a critical strain. At high crack velocity, strain accumulation decreases. In turn, higher crack driving force would be necessary for attaining the local fracture condition, thereby plastic zone size and plastic work being increased. In fact, the analysis under the identical value but with different crack velocity, Fig.23, has shown that the plastic work increases with increasing crack velocity, which is qualitatively consistent with Fig.18. Reason why the size of plastic deformation was so different and so was the plastic work between the DWTT and the full-scale burst test could be explained by this mechanism. Shim et al. conducted instrumented DWTT and evaluated the dependence of CTOA resistance value on crack velocity [14]. They obtained a

Copyright 2012 by ASME

decreasing CTOA value with increasing crack velocity, adverse tendency with the present results. This might be because their crack velocity range was limited, less than 100m/s, and the influence of plastic wave propagation does not make considerable effect. The present numerical analysis, however, is different from the pipe burst in the sense that the model is a plate in contrast to the cylindrical shape of a pipe. To reproduce as close as possible a pipe burst, bi-axial stress was applied in the model, simulating hoop as well as axial stress with ratio 1: 12 . However, the calculated plastic strain components were different from those of the full-scale burst test; the axial strain was comparable to the hoop strain in the full-scale burst in contrast to much smaller axial strain, , in the model, despite the assumed bi-axial stress. This might be because significant opening-up deformation of a pipe wall might produce axial strain. This kind of out-of-plane deformation was not reproduced by the model. However, the suppression of the plastic deformation at high crack velocity is essential both in the model and the pipe burst. In the present study, a crack resistance curve was constructed for X65 linepipe steels, irrespective of the steels tested. Further study must be made for developing a universal crack resistance curve. Measurement of plastic strains in fullscale burst tests is definitely important, in this regards. The resistance curve evaluated by a procedure like of the present study might be modified to conventional models, such as BTC and HLP. CONCLUSIONS (1) Measurements of distribution of plastic strains near a crack in DWTT, together with a stress-strain curve, enabled the estimation of plastic work. (2) Plastic work associated with crack propagation in the DWTT was nearly constant at 10mm to 30mm crack propagation from the precrack-tip. The plastic work at this location agreed with the absorbed energy after maximum load, that is, crack propagation energy. (3) Plastic strains were measured near a propagating crack in full-scale burst tested pipes. The plastic zone extended beyond 200 mm from the crack. Plastic work estimated from the distribution of the plastic strains was much larger than that of the DWTT. (4) The plastic work can be regarded as crack resistance for ductile crack propagation and plotted against crack velocity. A unique curve was derived for the tested X65 linepipe steels. (5) Elasto-plastic dynamic finite element analysis indicated that the accumulation of plastic strain at a crack-tip decreased with increasing crack velocity. This behavior can be understood as a result of the competition between plastic wave propagation and crack propagation. In turn, higher crack driving force is necessary for maintaining local fracture strain condition at the crack-tip, thereby

plastic deformation and plastic work being increased near the crack-tip. (6) Generalization of the crack resistance curve needs further experimental as well as numerical analyses. Measurement of plastic deformation near a crack in full-scale burst tests, especially for a long crack under steady state condition, is definitely important for this end. REFERENCES [1] R.J. Eiber, T.A. Bubenik, W.A. Maxey, Fracture control technology for natural gas pipelines, Technical Report PRCI PR-39113, NG-18, Battelle Memorial Institute, 1993. [2] H. Makino, T. Inoue, S. Endo, T. Kubo, T. Matsumoto, Simulation method for crack propagation and arrest of shear fracture in natural gas transmission pipelines, Pipe Dreamers Conf. Yokohama, Nov. 2002. [3] R.M. Andrew, N. Millwood, A.D. Batte, B. J. Lowesmith, The fracture arrest behavior of 914mm diameter X100 grade steel pipelines, Int. Pipeline Conf. Oct. 2004, Calgary, Canada, IPC04-0596. [4] H. Makino, I. Takeuchi, R. Higuchi, Fracture propagation and arrest in high-pressure gas transmission pipeline by ultra high strength line pipes, 7th Int. Pipeline Conf. Calgary Sept. 2008. [5] K. Misawa, Y. Imai, S. Aihara, A new model for dynamic crack propagation and arrest in gas pipelines, 8th Int. Pipeline Conf. Sept. 2010, Calgary, Canada, IPC2010-31475. [6] S. Aihara, H.I. Lange, K. Misawa, Y. Imai, Y. Sedei, E. Ostby, C. Thaulow, Full-scale burst test of hydrogen gas X65 pipelines, ibid, IPC2010-31235. [7] T. Namegawa, S. Aihara, K. Shibanuma, A. Takeshita, T. Sakimoto, S. Igi, Fractographic and FEM analyses of dropweight tear test of linepipe steel, ISOPE 2012, Rhodes, Greece, (accepted). [8] S.J. Garwood, Effect of specimen geometry on crack growth resistance, ASTM STP 677, pp.511-532, 1979. [9] Y. Imai, Influence of crack velocity and microstructure on fast ductile fracture resistance for high pressure gas pipeline steels, Masters Thesis, Feb. 2010, The University of Tokyo. [10] S. Aihara, E. Ostby, H.I. Lange, K. Misawa, Y. Imai, C. Thaulow, Burst tests for high-pressure hydrogen gas line pipes, 7th Int. Pipeline Conf. Sept. 2008, Calgary, Canada, IPC2008-64166. [11] SIMULIA, Abaqus Analysis User's Manual Version 6.10, Dassault Systemes, 2011. [12] W.J. Drugan, X.Y. Chen, "Plane strain elastic-ideally plastic crack fields for mode I quasistatic growth at large-scale yielding-I. A new family of analytical solutions", J. Mech. Phys. Solids, Vol.37, No.1, pp.1-26, 1989. [13] J.D. Achenbach, M.F. Kanninen, C.H. Popelar, Crack-tip fields for fast fracture of an elastic-plastic material, J. Mech. Phys. Solids, Vol.29, No.3, pp.211-225, 1981. [14] D.J. Shim, G. Wilkowski, D.M. Duan and J Zhou, "Effect of fracture speed on ductile fracture resistance - Part 1: Experimental", 8th Int. Pipeline Conf. Sept. 2010, Calgary, Canada, IPC2010-31310.

10

Copyright 2012 by ASME

You might also like