Professional Documents
Culture Documents
Zhu, Xian-Kui, Leis, Brian N (2006) - Application of Constraint Corrected J-R Curves To Fracture Analysis of Pipelines
Zhu, Xian-Kui, Leis, Brian N (2006) - Application of Constraint Corrected J-R Curves To Fracture Analysis of Pipelines
Keywords: crack-tip constraint, J-R curve, SENB, SENT, X80, J-A2 solution
Introduction the CTOD resistance curves for X52 pipeline steel. Hippert et al.
关7兴 reported experimental data of J-R curves and indicated the
Extensive experiments 关1–3兴 have indicated that the fracture
constraint effect on the J-R curves for X70 pipeline steel. Most
resistance curves against ductile crack growth, i.e., the J-R curves,
recently, Shen et al. 关8兴 published the detailed experimental results
depend on constraint levels at the crack tip. Fracture constraint
of J-R curves using a set of SENB and SENT specimens and
develops because of differences in loading and geometry configu-
demonstrated significant constraint effect on the J-R curves for
rations. The constraint dependence of J-R curves is an issue for
X80 and X100 pipeline steels. Pavankumar et al. 关9兴 performed
transferability of the experimental J-R curves measured using
full-scale fracture testing of J-R curves for cracked pipes with
standard laboratory specimens to actual cracked components.
different circumferential through wall cracks in four point bending
Standard specimens for fracture toughness testing, such as the
load, and their experimental results are also geometry dependent.
single edge notched bend 共SENB兲 and compact tension 共CT兲
Structural integrity assessment of pipelines is a key issue for the
specimens with deep cracks as specified in ASTM E 1820-01 关4兴,
design and operation of onshore and offshore oil and gas facilities.
have strict size requirements to ensure high constraint levels at the
Conventional failure assessment procedures applicable to pipe-
crack tip, while the nonstandard specimens or real flawed compo-
lines containing service-induced defects generally employ simpli-
nents with surface cracks usually involve low constraint levels. In
fied failure criteria that were empirically derived from experimen-
general, the application of J-R curves from high-constraint speci- tal analysis. Material failure 共leakage or sudden rupture兲 in a
mens to low-constraint geometry introduces a degree of conserva- flawed pipeline is preceded in ductile steels by large amounts of
tism into design analysis. Conversely, for some special cases, frac- slow stable crack growth until a critical crack size is reached.
ture toughness data are obtained from relatively low constraint Under sustained ductile tearing of a crack, a large increase in the
specimens and used in high constraint applications, which could
loading-carrying capacity of pipeline steels, as characterized by J-
result in a nonconservative design. A more detailed review on this
R curves, is possible beyond the limits by the conventional analy-
subject can be found in Chao and Zhu 关5兴.
sis 关7兴. Realistic elastic-plastic fracture assessments of pipelines
In recent years, the study of constraint effect on fracture resis-
rely on the direct application of J-R curves measured with small
tances for pipeline steels has attracted much attention. Pisarski
laboratory specimens. Therefore, the constraint dependence of J-R
and Wignall 关6兴 presented experimental data as CTOD fracture
curves needs to be understood, and a viable procedure needs to be
toughness resistance curves and showed the constraint effect on
developed to transfer the fracture resistance data measured with
small laboratory specimens to those for actual components and
Contributed by the Pressure Vessels and Piping Division of ASME for publication structures.
in the JOURNAL OF PRESSURE VESSEL TECHNOLOGY. Manuscript received July 26, 2005; Much theoretical and numerical work has been done to quantify
final manuscript received December 6, 2005. Review conducted by Kenneth K. fracture constraint at the crack tip in both two- and three-
Yoon. Paper presented at the 2005 ASME Pressure Vessels and Piping Conference
共PVP2005 “Outstanding Technical Paper”兲, July 17–21, 2005, Denver, Colorado, dimensional specimens and structural configurations 关10–15兴. Due
USA. to the loss of single-parameter J dominance, several two-
Journal of Pressure Vessel Technology Copyright © 2006 by ASME NOVEMBER 2006, Vol. 128 / 581
parameter fracture theories have been proposed. One is the J-Q reduction of area is 68.3%. In the elastic-plastic fracture analysis,
theory proposed by O’Dowd and Shih 关10,11兴, the other is the it is usual to fit the true stress-strain curve in Fig. 1 by the form of
J-A2 three-term solution proposed by Yang and Chao 关12,13兴, Ramberg-Osgood power-law relation:
冉 冊
where Q and A2 are constraint parameters. The comparisons of
these two solutions can be found in Chao and Zhu 关14兴 and Kim et n
= +␣ 共1兲
al. 关15兴. Based on experimental data, finite element analysis 0 0 0
共FEA兲 results, and the J-A2 solution, a new methodology has been
where the reference stress 0 is taken as the 0.2% offset yield
developed by Chao and Zhu 关5,16兴 for constructing a constraint
stress, the reference strain 0 = 0 / E, and Young’s modulus E
corrected J-R curve for ductile crack growth, using A2 as the
= 207 GPa. As evident in Fig. 1, the use of the hardening param-
constraint parameter. Recently, Lam et al. 关17兴 have applied this
eter ␣ = 1.07 and n = 13 leads to a best fit of the test data.
methodology to characterize the constraint effect on the fracture
resistance curves for carbon steel A285. This paper further ex- Specimen Geometry. The SENB and SENT specimens were
tends this methodology to determine a constraint corrected J-R cut from the pipe section in LC orientation and used to measure
curve for an X80 pipeline steel. J-R curves. Dimensions of the SENB specimens recommended by
Based on J-R curves measured for the X80 pipeline steel by a ASTM E 1820 were used, except for the initial crack length,
set of SENB specimens, the fracture initiation toughness at the which was varied to develop different levels of crack-tip con-
crack extension of 0.2 mm and the J-integral values at the crack straint. All SENB specimens were loaded in three-point bending
extension of 1.0 mm have been determined. The values of con- with a span of 92 mm, after precracking by fatigue according to
straint parameter A2 have been obtained by matching the J-A2 the procedure in ASTM E 1820. Specimens were side grooved
solution with the FEA results for each specimen. 共Note that the using a Charpy cutter to a total thickness reduction of 20%, in an
modified J-A2 four-term solution 关18兴 should be used for bending attempt to develop plane strain conditions along the crack front.
specimens with deep cracks.兲 Using the power-law relationship The specimen width W was 23 mm with initial crack length a
suggested by ASTM E 1820, a constraint corrected J-R curve has varied producing ratios of a / W between 0.24 and 0.64. This in-
been formulated for the specific X80 steel considered as a func- terval represents shallow to deep cracks and simulates different
tion of the constraint parameter A2 and crack extension ⌬a. Cor- fracture constraint levels at the crack tip. The uncracked ligament
respondence between the predicted J-R curves and the experimen- length b = W − a and specimen thickness B = W / 2 = 11.5 mm. All
tal data will be evaluated for both SENB and SENT specimens. specimens were tested at room temperature 共about 20° C兲.
Application of this method then will be demonstrated in reference
to a pipeline with a deep axial surface crack. From the predicted Experimental J-R Curves. The J-R curves of the X80 pipeline
J-R curve and crack drive force obtained by FEA, the failure steel were measured by Shen et al. 关8兴 for both SENB and SENT
pressures of the pipeline at crack initiation and instability are de- specimens. For the SENB specimens, determinations of J-R
termined, and compared with those obtained from plastic collapse curves follow the standard procedures of ASTM standard E 1820.
analysis. After testing, the specimens were heat tinted and then broken in
liquid nitrogen. The initial and final crack lengths were measured
on the fracture surface by the nine-point technique as described in
Experimental Results ASTM E 1820. The criterion for uniform crack extension given in
ASTM E 1820 共which requires that none of the nine physical
Material Properties. The material considered was a high measurements differ by more than 5% from the average physical
strength pipeline steel, API X80, i.e., N550 supplied by crack size兲 was not met for the final crack length, primarily be-
TransCanada Pipelines Ltd. 共TCPL兲 as detailed in Ref. 关8兴. As cause splitting occurred during the tests. For some specimens with
detailed therein, all specimens were cut and machined from a pipe severe splitting, the difference among the nine physical measure-
section with outside diameter of 1219 mm 共48 in. 兲 and wall ments for the final crack length was as high as 40%. However,
thickness of 12.7 mm 共0.5 in. 兲 The tensile properties were mea- splitting was not observed for crack extension ⌬a ⱕ 0.2 mm. Fig-
sured using full-section 共25⫻ 11.5 mm2兲 flattened samples ori- ure 2 shows the experimental J-R curves for the X80 SENB
ented in the longitudinal direction. The true stress-strain curve is specimens.
illustrated in Fig. 1. The tensile test results show that the 0.2% The initiation toughness 共Ji兲 used to characterize fracture resis-
offset yield stress of the X80 is 570 MPa, the 0.5% total yield tance at the onset of stable ductile crack growth can be defined on
stress is 576 MPa, the ultimate tensile stress is 675 MPa, the elon- the J-R curve at the critical stretch zone width 共⌬ai兲, which can be
gation for the 1 in. 共25 mm兲 gage length is 42.2%, and the final measured on the fracture surface using a scanning electron micro-
共k兲
where the stress angular functions ˜ij 共兲 共k = 1, 2, 3兲 and the
共3兲 共0兲
0
= A1 冋冉 冊r
L
s1
共1兲
˜ 共0兲 + A2 冉冊
r
L
s2
共2兲
˜ 共0兲 + A22 冉冊
r
L
s3
共3兲
˜ 共0兲 册
stress power exponents sk 共s1 ⬍ s2 ⬍ s3兲 depend only on the strain
6Mr
hardening exponent n and are independent of the other material − 共5兲
constants 共i.e., ␣, 0, and 0兲 or the parameter A2. The values of 0b 3
˜共k兲
ij 共兲 and sk are tabulated in Ref. 关21兴. In Eq. 共3兲, L is a charac- where M is the global bending moment. The modified J-A2 solu-
teristic length parameter, with L = 1 mm being adopted in this tion still only involves the two parameters, namely the applied
work. The parameters A1 and s1 are related to the HRR singularity load 共J and M兲 and constraint parameter 共A2兲.
field 关22,23兴 by
冉 冊
Determination of Constraint Parameter A2. The constraint
−s1
J 1 parameter A2 is usually determined by matching the opening
A1 = , s1 = − 共4兲
␣ 0 0I nL n+1 stress, , from the J-A2 solution with the FEA result at = 0ⴰ and
r / 共J / 0兲 ⬇ 2. Figure 6 depicts the variations of the constraint pa-
and s3 = 2s2 − s1 for n ⱖ 3. The constraint parameter A2 is an un-
known constant and can be determined by matching the opening rameter A2 with the applied loading J for the SENB specimens
stress from the J-A2 solution with the FEA result at r / 共J / 0兲 with a / W = 0.24, 0.42, and 0.64. The parameter A2 in Fig. 6共a兲 is
determined by the J-A2 solution or Eq. 共3兲, while the parameter A2
= 1 – 2, for example.
in Fig. 6共b兲 for the modified solution is determined by the modi-
The Modified J-A2 Solution. For deep crack bending speci- fied J-A2 solution or Eq. 共5兲. Comparison shows that the A2 values
mens under LSY or fully plastic deformation, it has been shown determined using Eqs. 共3兲 and 共5兲 are almost the same and con-
that the global bending stress significantly affects the crack-tip stant for all loading for the shallow crack 共a / W = 0.24兲, but are
field. As a result, the J-A2 three-term solution, i.e., Eq. 共3兲, fails to different for the two deeper cracks under LSY. Without the global
correctly describe the crack-tip field. To eliminate the influence of bending influence, A2 is theoretically a constant independent of
冉 冊 C2共A2兲 A2. The valid range of A2, based on past studies, is between −1
⌬a
J共⌬a,A2兲 = C1共A2兲 共6兲 and 0, as the crack tip constraint varies from low to high. Solving
k C1 and C2 for a series of A2 values within this range provides the
functional dependencies of C1 and C2 with respect to A2. Finally,
where k = 1 mm and the coefficients C1共A2兲 and C2共A2兲 are unde-
least-squares regression analysis provides the desired functional
termined functions of A2. It should be noted that C1 and C2 are
forms of C1 and C2 in terms of the constraint parameter A2.
constants in the original ASTM E 1820 formulation. Equation 共6兲
For a given material, once the expression for the constraint-
extends the current ASTM J-R curve concept, J共⌬a兲, to a con-
corrected J-R curve, or Eq. 共6兲, is obtained, the J-R curve can be
straint corrected J-R curve, J共⌬a , A2兲. This requires determining accurately predicted for any specific cracked geometry 共e.g., non-
the functional dependencies of C1 and C2 on the constraint param- standard specimens or actual structural components兲, provided
eter A2. Once the functional forms of C1共A2兲 and C2共A2兲 are ob- that the constraint parameter A2 is known for that cracked geom-
tained, a family of constraint-corrected J-R curves is completely etry.
determined.
As indicated above, Eq. 共6兲 contains two unknown variables, C1 Construction of Constraint-Corrected J-R Curve. Using the
and C2 for a given A2; consequently, two equations are needed to SENB test data shown in Fig. 2 and the method described above,
solve for these values. Chao and Zhu 关5兴 proposed a procedure to a constraint corrected J-R curve in terms of A2 can be constructed
determine the functions C1 and C2 in reference to the initiation for X80 pipeline steel. Two equations are set up to solve for C1
toughness and the ductile tearing modulus defined in ASTM E and C2 at J = J0.2 mm and J = J1.0 mm. Using the 共modified兲 J-A2
1820. A more general approach is developed here. Basically, the solution at J0.2 mm, the value of A2 for each SENB specimen is
two equations can be established at any two points of crack ex- determined. Figure 7 plots the J0.2 mm versus A2 and J1.0 mm ver-
tension for the experimental J-R curves. The first point can be sus A2 relations. From this figure, two linear equations between
chosen at crack extension ⌬a1, where the crack initiates, and the J0.2 mm and A2, and J1.0 mm and A2, are fitted as follows:
J0.2 mm = − 1505.4A2
共9兲 Application to Failure Predictions of Cracked Pipeline
J1.0 mm = − 4417.1A2
FEA Simulation and Constraint Analysis. A pipeline with an
where the J integral has unit in kJ/ m2. It is noted that a constant outside diameter 762 mm and a thickness 23 mm under internal
J0.2 mm = 337.3 kJ/ m2 was used in our preliminary work 关24兴. pressure is considered. For purposes of analysis, it is assumed that
Substitution of Eq. 共9兲 into Eq. 共8兲 leads to: the pipeline contains a very long axial surface crack on its exterior
with the crack depth being 50% of the wall thickness. Since the
C10 . 2C2 = − 1505.4A2 crack is very long, plane strain deformation can be reasonably
共10兲 assumed for the cracked pipe. As a result and in light of symme-
C11 . 0C2 = − 4417.1A2 try, one half of the circular cross-section is modeled using the
FEA mesh shown in Fig. 10. In this FEA model, a very fine mesh
From Eq. 共10兲, the functions of C1共A2兲 and C2共A2兲 can be ap- is generated near the crack tip with the smallest element size of
proximated as 0.003 mm, with coarse mesh used elsewhere. The mesh involves
3107 nodes and 944 eight-node plane strain isoparametric ele-
C1共A2兲 = − 4417.1A2 ments with reduced integration. The boundary conditions reflect
共11兲 symmetry and the only applied load is internal pressure.
C2共A2兲 = 0.668 Figure 11 shows the distribution of crack opening stress deter-
mined from the FEA calculation and the J-A2 solution 共Eq. 共3兲兲
Finally, substituting Eq. 共11兲 into Eq. 共6兲, we obtain the constraint- along the distance from the crack tip at different deformation lev-
corrected J-R curve for X80 pipeline steel as follows: els from SSY to LSY for the X80 pipe with a surface crack of
a / t = 0.5, where the constraint parameter A2 is taken as −0.25 for
J共⌬a,A2兲 = − 4417.1A2 冉 冊⌬a
1 mm
0.668
, 共12兲
all loading. Note that the HRR solution is also included in Fig. 11.
It is apparent that the global bending stress has insignificant influ-
ence on the crack tip, and the J-A2 three-term solution agrees with
where the J integral has unit of kJ/ m2. Equation 共12兲 indicates the FEA results at all deformation levels, whereas the HRR field
that the constraint corrected J-R curve is a function of A2 and ⌬a. does not fit the FEA results even under SSY. This indicates that
If A2 is known for a specific geometry, the J-R curve can be easily the J-A2 solution can be used to characterize the crack tip field for
predicted from Eq. 共12兲. the surface crack in this X80 pipeline.
Validation of Predicted J-R curves. For the SENB specimens Predicted J-R Curve of Cracked Pipe. Replacing the value of
with a / W = 0.24, 0.42, and 0.64, the A2 is determined at J0.2 mm −0.25 for A2 in Eq. 共12兲 of the constraint-corrected J-R curve
= 337.3 kJ/ m2 as −0.27, −0.23, and −0.18, respectively. Thereaf- yields the predicted J-R curve for the X80 pipe with a long axial
ter, the J-R curve can be predicted from Eq. 共12兲. Figure 8 com- surface crack of a / t = 0.5 as follows:
冉 冊
pares the predicted J-R curves with the experimental J-R curves
for these three SENB specimens. Likewise, Fig. 9 compares the ⌬a 0.668
predicted J-R curves with the experimental J-R curves for the J共⌬a兲 = 1104.25 共13兲
1 mm
SENT specimens with a / W = 0.25, 0.41, and 0.57, respectively.
Comparisons show that the predicted J-R curves from Eq. 共12兲 where the J integral is in kJ/ m2. This J-R curve is plotted in Fig.
match well with the experimental data for both SENB and SENT 12 and compared with those for the SENB specimens.
specimens and provide better predictions than the constraint- It indicates that the J-R curve for the cracked pipe is much
corrected J-R curves using a constant value of J0.2 mm in our pre- higher than that for the ASTM standard SENB specimen with
liminary work 关24兴. Therefore, the constraint-corrected J-R curve a / W = 0.64 or 0.5, and may coincide with that for the SENB speci-
or Eq. 共12兲 can be effectively used to predict the J-R curve for any men with a / W = 0.35. Therefore, using the J-R curve from the
specimen or actual component, provided that the material is the ASTM standard specimen could overly underestimate the fracture
considered X80 pipeline steel and the constraint parameter A2 is resistance and the limit load of the X80 pipeline with surface
known a priori. cracks.
再 冋 冉 冊册冎
for blunt defects using the following equation:
ditions, Eq. 共15兲 predicts the burst pressure as pd = 21 MPa at plas-
2t d L tic collapse. Under LSY conditions in ductile steels, initially crack
pd = uts 1− 1 − exp − 0.157 共14兲
D t 冑R共t − d兲 tips blunt, so Eq. 共15兲 should provide a reasonable estimate of the
limit pressure at plastic collapse for the X80 pipeline under con-
where uts is the ultimate tensile stress, D ⫽ 2R is the average sideration. The next paragraphs assess the consistency between
diameter of the pipe, t is the wall thickness, d is the defect depth, this “macro” model and the behavior characterized by fracture
and L is the defect length. PCORRC is the ultimate tensile stress mechanics.
共UTS兲 based plastic collapse criterion for blunt defects derived
independent of fracture mechanics analysis. Therefore, the limit Elastic-Plastic Fracture Failure. In the elastic-plastic fracture
pressure determined by Eq. 共14兲 is a plastic collapse failure load. mechanics, the failure criterion for a crack is crack driving force J
For very long defects, Eq. 共14兲 is simplified as reaches its fracture resistance JR, namely,
J = JR 共16兲
Fig. 11 Distribution of the opening stress determined from the Fig. 12 Predicted J-R curve for X80 pipe with a surface crack
FEA and J-A2 solution along the distance from the crack tip and compared with those for SENB specimens