ADVANCES IN CRACK ASSESSMENT FOR PIPELINE INTEGRITY

David Katz1, Ming Gao2, Sergio Limon2 and Ravi Krishnamurthy3

1
Williams Gas Pipeline West, 295 Chipeta Way, MS-3P1, Salt Lake City, Utah 84158-0900, USA
2
PII Pipeline Solutions, GE Energy, 2707 North Loop West, Houston, TX 77008, USA
3
Blade Energy Partners, 11111 Katy Freeway, Suite 910, Houston, TX 77079, USA

ABSTRACT

Recently, the Office of Pipeline Safety (OPS) of the United States of America issued an advisory notice to all US
pipeline owners and operators to consider stress corrosion cracking (SCC) as a safety risk to their pipeline and to
include SCC assessment and mitigative measures in their integrity management plans. If a pipeline is susceptible to
SCC, an appropriate in-line inspection technology and a hydrostatic testing program are two main options to identify
and expose SCC. Fracture mechanics (FM) assessments are then recommended to estimate where in the system an
SCC immediate threat might occur and to quantify the life cycle of the pipeline. In this paper, the benefits and
limitations of ILI inspection and hydrostatic testing are critically reviewed. Advances in fracture mechanics
methodologies for SCC evaluation in pipelines are summarized and specific issues associated with application of FM to
pipelines are also discussed.

1 INTRODUCTION

There is an increasing concern of SCC in the United States of America due to recent failures, not
only in natural gas pipelines but also in hazardous liquid transporting pipelines. Because of the concern,
the Office of Pipeline Safety (OPS, [1]) issued an advisory notice in October 2003 to all US owners and
operators to evaluate their systems for the presence of risk factors associated with high-pH (9-11) or near-
neutral pH (6-8) SCC. In accordance with ASME B31.8S (ASME [2]), a pipeline segment should be
considered susceptible to high-pH SCC if it is (1) operated at a pressure above 60% SMYS, (2) located
<10 miles down stream from pump or compressor station, (3) operated at a temperature exceeding 38°C
(100ºF) and (4) is more than 10 years old and protected with a coating other than fusion epoxy bonded. For
near-neutral pH SCC, the same criteria can be applied with the exclusion of the temperature criterion
(NACE SCCDA [3]). If conditions for SCC are present based on the evaluation, an operator should
prioritize the application of in-line inspection, hydrostatic testing and other forms of integrity verification
(OPS [1]). This should be followed by a fracture mechanics assessment to quantify the life cycle of the
pipeline and to take appropriate actions to mitigate areas of concern.
As one of the actions responding to OPS’ notice, a critical review of the benefits and limitations of
the in-line ultrasonic crack detection (USCD) tool as compared with the hydrostatic testing is given.
Advances in fracture mechanics assessment methodologies, particularly, the elastic-plastic fracture
mechanics based two-criteria failure assessment diagram approach and applications to pipeline SCC
evaluation are reviewed.

2 CRITICAL REVIEW–IN LINE INSPECTION vs HYDROSTATIC TESTING

Hydrostatic testing of existing pipelines has been widely used to demonstrate or revalidate the
pipeline integrity and serviceability (Keifner, [4]). Both field experience and full-scale laboratory tests
have revealed the benefits and limitations of hydrotesting. On the other hand, numerous in-line inspection
(ILI) practices showed that the appropriate ILI technologies are often superior alternatives, for example, the

1
magnetic flux leakage (MFL) tool for the detection of metal loss caused by corrosion. Recent evolution of
in-line inspection technologies has provided new tools for detecting much smaller seam weld anomalies
than those found by hydrostatic testing of up to 110% SMYS (Grimes [5]).
For SCC detection, the shear wave ultrasonics employed by the USCD tool have successfully been
used since its introduction in 1994 for more than 7000km of crack inspection (Marreck [6], Wolf [7,9],
Uzelac [8]). It is a proven technology capable of reliably detecting and sizing crack-like features larger
than 30mm (1.2 inch) in length and 1mm (0.04 inch) in depth, including incipient of SCC which is
significantly better than hydrotesting for crack detection in the critical regime of SCC, Figure 1.

1 0 0%

2 6" O D , 0.2 8" W T , A P I 5 L X 5 2 G rad e

7 5%
Crack Depth (% wt)

5 0%
M AO P = 72 % S Y M S

H yd ro te s t @ 1 .5 2 M AO P = 1 10 % Y S

2 5%

U S C D T O O L T H R E S H O L D 1 .2" x 0 .0 4 "

0%
0 2 4 6 8 10 12 14
C rac k L e ng th , (in )

Figure 1: ILI and hydrostatic testing comparison, 26” OD, 0.281” wt, API 5L X52 grade steel.

Another fundamental difference in crack detection and assessment between these two technologies
is that the ILI provides detailed information on crack location, size and distribution along the pipeline
which can be used for the development of a comprehensive integrity management plan (Marreck [6]),
whereas hydrostatic testing is a type of snapshot testing that removes all cracks greater than critical size at
the test pressure but provides no information on the remaining sub-critical cracks. Figure 2 illustrates how
the cracks were assessed using API 579-2000 Level-III (i.e., material specific) FAD for Williams’ 16-inch
natural gas transporting pipeline (Katz [10, 11]). An excavation, monitoring plan and re-inspection interval
were developed based on an assumed but commonly used crack growth rate to manage the integrity of the
pipeline. No comparable integrity plan could be developed for this line after subjecting to hydrostatic
testing in 1993 (Katz [10]).
F A D , A P I 5 7 9 L e v e l 3 , Y S = 7 2 .2 k s i, U T S = 8 9 .6 k s i
1 .2

1
U n a c c e p ta b le
K r C u rv e fo r L e v e l 3
0 .8 M A O P
1 .3 9 M A O P
Kr

0 .6

0 .4 A c c e p ta b le

0 .2

0
0 0 .2 0 .4 0 .6 0 .8 1 1 .2 1 .4

L r

Figure 2: API 579 Level-III FAD showing that all anomalies are acceptable
There are concerns about cracks possibly not being detected by the in-line inspection tools. Field
experience has shown that the probability of non-detection is very low. The risk of non-detection is

2
minimized with an appropriate validation excavation program and FM based integrity plan. In the same
context, one must also recognized that hydrostatic testing is not foolproof either. Because hydrotesting can
leave behind cracks that could be detected by in-line inspection, the use of hydrotest often demonstrates
serviceability for only a short period of time if a crack-growth mechanism exists (Kiefner [4]).

3 ADVANCES IN FRACTURE MECHANICS METHOD FOR PIPELINE CRACK ASSESSMENT

Oil and gas transmission pipelines have a good safety records even though SCC is now becoming
an increasing concern. This is because the nature and behavior of various defects in pipelines have been the
subject of considerable study over the past 40 years. Particularly, for crack or crack like defect assessments,
a two-parameter fracture mechanics based approach, known as NG-18 equations (eqs. 1-4), was introduced
in late 60s (Hahn [12]) and early 70s (Maxey [13] and Kiefner [14]) using the strip yield model (Dugdale
[15]).

Kc2 = (8c(σfs)2/π) LnSec [πMpσΗ/(2σfs)] (1)

Kc2 = 12CvE/Ac (2)
12CvπE/[8Acc(σfs)2] = LnSec [πMpσΗ/(2σfs)] (3)
Mp = [(1-d/t)(MT)-1]/(1-d/t) (4)

Where Kc = fracture toughness, Cv = upper shelf Charpy impact energy, C = half effective flaw length, E =
elastic modulus, σΗ = nominal hoop stress due to internal pressure, σfs = flow stress, Ac = the cross-
sectional area of the Charpy impact specimen, Mp = stress magnification factor, d = flaw depth, t = wall
thickness and MT = Folias bulging factor.
The NG-18 LnSecant method has been widely used for ERW seam weld defects, railroad fatigue
crack assessment, and has recently been recommended for material toughness evaluation for low frequency
ERW and LAP welded longitudinal seam evaluation (Baker [16]). Generally, the NG-18 LnSecant
methods are considered to be very conservative, particularly, for fatigue life predictions. However recent
experience in application of this method to SCC found that predictions of failure pressure and critical size
could be non-conservative due to the use of overly estimated fracture toughness values for the assessment.
The fracture toughness Kc in the NG-18 equations is calculated from an upper-shelf Charpy
impact energy value using an empirical relation (eq. 2) developed from full-scale burst tests of line pipes
that contained mechanically machined flaws. Using this empirical relation, the calculated Kc value is
found to be two or three times higher than the actual measured value from pre-cracked specimen of the
same material. As shown in Figure 3, a significant difference is observed between the calculated Kc values
from eq. 2 and the actual measured KJMAT (Anderson [17], Jaske [18]) and those estimated from other
empirical relations (Thorby [19], Wilkowski [20], Leis [21]). Consequence, the predicted failure pressure
and crack size would be significantly higher and larger, as compared to the respective values if the actual
measured KJMAT is used. This influence is shown in Figure 4.
Another drawback of the NG-18 equations is that the stress magnification factor, Mp, (eq. 4)
exhibits a singularity at the point of d = t. This provides inconsistent results when the crack approaches a
through-wall crack configuration. This inconsistency will result in conservatism in the computation of the
failure pressure when d/t > 0.5. The singularity nature of eq. 4 combined with the overly estimating
fracture toughness for cracked structures, results in a large uncertainty in predicting critical crack sizes and
failure pressures.

3
 450

400

Kiefner's Relationship
350 0.5
Kc = (12ECv/A)
0.5 0.5
= 18.97 (Cv/A) ksi in
300

250
Kc

200 Thorby et al., [19], low and From Leis and Burst [21]
0.5 0.5
medium strength steels, Kc = 6.1(Cv/A) ksi in
0.534 0.5
Kc = 5.92(Cv/A) ksi in
150

PII Data
100

Wilkowski et al., nuclear

piping steels [20],
50
0.5 0.5
Kc = 5.45(Cv/A) ksi in

0
0 10 20 30 40 50 60 70

CVN, ft-lb

Figure 3: A comparison of Kc between NG-18 method, actually measured and other reported values
2000

Elliptical C-Equiv, D=22.000 inch, t=0.334 inch, SMYS = 46,000 psi

1800 CVN = 18.00 ft-lb, CVN Area = 0.124^2

Kc = 228 ksi in0.5 (calculated from 18 ft-lb using eq. 2 [15])

1600
KJc = 120 ksi in0.5, actual measured [22]

1400
Failure Pressure, psi

1200

10%
1000

800

600 10%

400
100% ( Through-Wall)

100% ( Through-Wall)
200
90%
90%
0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Crack Length, Inches

Figure 4: The influence of Kc on the predicted failure pressure and crack size

It is noted that several years after NG-18 methods were introduced, the two-criteria failure
assessment diagram concept was further developed by Dowgling et al [22] and Harrison et al [23], which
describes the interaction between fracture and plastic collapse. This FAD approach forms a basis for
industry practice documents (PD 6493) and standards (BS7910:1999 [24], API 579-2000 [25]) that can be
applied to assessment of cracks or crack-like flaws in pipelines. In addition to the Rainbow Pipeline
system (Krishnamurthy [26]), limited experience gained from the Williams pipelines has shown that the
FAD method provides conservative but consistent predictions for fitness-for-purpose (FFP) evaluation and
more accurate results for failure analysis if a material specific FAD is utilized. For example, a rupture
failure mode is predicted when the defect size is about 6.4-inch long and 71% deep in a 26-inch OD API 5L
X52 pipe based on the tearing instability analysis, Figure 5. This prediction is consistent with the field
observations. However, for the same operating and pipe conditions, a leak failure mode would have been
predicted by NG-18 equations. Moreover, API 579 provides reliable results for surface defects d/t > 0.5.

4
 2 .5

2 6 " O D , 0 .2 8 1 " w t, A P I 5 L X 5 2
M O P = 630 psi
C r a c k D e p th = 0 .2 0 " (7 1 % W T ) 10"
2

7 .0 "

6 .4 "
6"
1 .5 5"

4"

Kr 3"

0 .5

0
0 0 .2 0 .4 0 .6 0 .8 1 1 .2

Lr

Figure 5: Ductile tearing analysis predicts a rupture mode for cracks longer than 5 inches

Finally, the predicted critical crack size between API 579 FAD approach and NG-18 method at
two pressures (i.e., 0.72% and 110% SMYS) is compared and shown in Figure 6. It is seen that the
difference in predictions is significant. The NG-18 method provides more conservative results for longer
cracks but the opposite for shorter cracks. This could be attributed to the NG-18 equations’ combined
effect of overly estimated fracture toughness and the singularity of Mp. Another difference between these
two methods is that API 579 predicts critical depth relatively insensitive to crack length when it gets longer
than 6-inches while NG-18 does not, Fig. 6. This difference appears to result from the fact that the NG-18
equations were originally derived from a through-wall crack configuration based on the Dugdale strip yield
model. The modification of the NG-18 equations for surface cracks was made by introducing a stress
magnification factor Mp for stress calculation without considering the change in stress intensity field at the
crack tip (Kiefner [14]). On the contrary, the API 579 FAD method includes both [25].
100%

75%
26” OD,
Crack Depth (%wt)

72% YS NG-18
0.281”WT
API 5L X52.
50% 72% SMYS API 579

110% YS API 579

25%
110% YS NG-18

0%
0 2 4 6 8 10 12 14
Crack Length, (in)

Figure 6: A Comparison of critical crack size prediction between API 579 FAD and NG-18 methods,

4 SUMMARY

A critical review of the benefits and limitations of crack detection tools (USCD) and hydrostatic testing for
SCC evaluation suggests that ILI inspection is the more appropriate integrity management methodology for
piggable pipelines. ILI inspection with a good probability of detection basis, followed by an appropriate
fracture mechanics analysis could result in a comprehensive pipeline management plans. Limited
experience in crack assessments using elastic-plastic fracture mechanics methods suggests that the two-
criteria FAD approach is more appropriate than NG-18 method. Industry standards API 579 and BS 7910
approach provides conservative but consistent results for FFP. This approach also provides opportunities
for more accurate high level FAD assessment using material specific and finite element analysis (FEA)
input data to meet various engineering purposes.

5
 5 REFERENCES

[1] Office of Pipeline Safety, “SCC Threat Advisory Notice,” October 2003
[2] American Society of Mechanical Engineer, “ASME B31.8S Supplement to B31.8,” ASME
International, New York, NY, 2002
[3] National Association of Corrosion Engineer, “NACE: External SCC-Direct Assessment” February 2004
[4] Kiefner, J., F., “Role of Hydrostatic Testing in Pipeline Integrity Assessment,” Northeast Pipeline
Integrity Workshop, Albany, NY, June 2001
[5] Grimes, K, “A Breakthrough in the Detection of Long Seam Weld Defects in Steel Pipelines,” Pipeline
Integrity International, Cramlington, UK, 1991
[6] Marreck, P., Martens, B., Krishnamurthy, R., Tozer, N., “Mobil Oil’s experience with
in-line detection and characterization of SCC”, Pipeline Pigging Conference, 1999.
[7] Wolf, T., “Pipeline Inspection with UltraScan®”, OIL GAS European Magazine, Vol. 2, 2001.
[8] Uzelac, N., “Reliable detection of SCC and other cracks in gas and liquid pipelines”,
Pipeline Integrity Congress 2000, Buenos Aires, Argentina.
[9] Wolf, T., “Ultrasonic Inspection is used most to detect and size cracks”, Pipeline & Gas
[10] Katz, D., Gao, M., Sherstan, R., Elboujdaini, M., Li, J., “Continuing Data Assessment of 16” Williams
Pipeline Inspected with the Recently Developed Ultrasonic Crack Detection Tool”, Proceedings of the
42nd Conference of Metallurgists, COM 2003, Vancouver, British Columbia, Canada, p.265.
[11] Katz, D., Rühle, U., Wolf, T., “16-inch Williams Gas Pipeline Inspected for SCC using PII’s New
Ultrasonic Tool,” IPC 2002, paper No:27055, Calgary, Canada, 2002.
[12] Hahn, G.T., Sarrate, M., Rosenfield, A. R., “Criteria for Crack Extension in Cylindrical Pressure
vessels” International Journal of Fracture Mechanics, Vol. 5, No.3, 1969, pp.187-210.
[13] Maxey, W. A., Kiefner, J. F., Eiber, R. J., Duffy, A. R.: “ Ductile Fracture Initiation, Propagation, and
Arrest in Cylindrical Vessels”, ASTP STP 514, pp. 71-83.
[14] Kiefner, J. F., Maxey, W. A., Eiber, R. J. and Duffy, A. R., ”Failure Stress Levels of Flaws in
Pressurized Cylinders”, ASTM STP 536, 1973, pp. 461-481.
[15] Dugdale, D.S., “ Yielding in Steel Sheets Containing Slits”, Journal of the mechanics and Physics of
Solids, Vol. 8, 1960, pp 100-104.
[16] Baker, M. Jr., “Low Frequency ERW and Lap Welded Longitudinal Seam Evaluation”, Final Report,
DOT OPS, TTO Number 5, November, 2003.
[17] Anderson & Associates, J-Integral Testing, Internal Report GE-PII, Houston, TX, 2003.
[18] Jaske, C. E., Vieth, P. H., Beavers, J. A., “Effect of Stress Corrosion Cracking on Integrity and
Remaining Life of Natural Gas Pipelines”, CORROSION96, Paper No. 255, NECA, 1996.
[19] Thorby, P.N., Ferguson, W.G., “The Fracture Toughness of HY60”, Materials Science and
Engineering, Vol. 22, 1976, pp. 117-184.
[20] Wilkowski, G.M., “Degraded Pipeline Program – Phase II, Semiannual Report, April 1986 –
September 1986”, NUREG/CR-4082, Vol. 5, Battelle’s Columbus Division, April 1987.
[21] Leis B. N., Brust, F. W., ”Ductile Fracture Properties of Selected Linepipe Steels”, NG-18 Report No.
204, Pipeline Research Committee of the American Gas Association, Inc., Washington, D.C. 1992.
[22] Dowling, A. R., Townley, C.H.A., “ The Effects of Defects on structural failures: A Two Two-Criteria
Approach”, International Journal of Pressure vessels and Piping, Volume 3, 1975, pp. 77-137.
[23] Harrison, R. P., Loosemore, K., Milne, I. “Assessment of the Integrity Structures Containing Defects”,
CEGB Report R/H/R6, Central Electricity Generating Board, UK, 1976.
[24] British Standard Institute, “Guide on Methods for Assessing the Acceptability of Flaws in Metallic
Structure BS 7910,” British Standard Institute, June 1999.
[25] American Petroleum Institute, “Fitness-for-Service API 579,” API Publishing Services, January 2000
[26] Krishnamurthy R. M, McDonald R.W., Marreck P. M., “ Proc. Of the 1st International Pipeline
Conference”, 1996, p. 495.