Ipc2012-90089-Research On Burst Tests of Pipeline With Spiral Weld Defects#Chen2012

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

Proceedings of the 2012 9th International Pipeline Conference

IPC2012
September 24-28, 2012, Calgary, Alberta, Canada

IPC2012-90089

RESEARCH ON BURST TESTS OF PIPELINE WITH SPIRAL WELD DEFECTS

Jian Chen Qingshan Feng


PetroChina Pipeline R&D Center PetroChina Pipeline Company
Langfang, Hebei, China Langfang, Hebei, China

Fuxiang Wang Hailiang Zhang Hancheng Song


PetroChina Pipeline R&D Center PetroChina Pipeline Company PetroChina Pipeline Company
Langfang, Hebei, China Langfang, Hebei, China Langfang, Hebei, China

ABSTRACT
On some old long distance transmission pipelines of NOMENCLATURE
PetroChina some severe spiral weld defects existed. These weld A - longitudinal area of metal loss
defects were found to be from a lack of penetration and lack of A( ,  ) - spiral correction factor
fusion during pipe manufacturing. To avoid possible failure
c - half length of the defect
accidents, tri-axis magnetic flux leakage (MFL) in-line
inspections were performed on some certain pipelines. A large D - pipe diameter
quantity of spiral weld defects were detected in the inspections d - maximum depth of the defect
and their sizes were reported, some of them with large depth K r - ratio of SIF to material toughness
and large length along the spiral weld seam. However, there is
no applicable engineering critical assessment (ECA) method
L - length of the defect
for this kind of defects up to now. Assessment methods M - Folias factor
provided in BS7910:2005 and other codes are only applicable Pb - predicted burst pressure
to assess flaws in cylinder oriented axially or circumferentially.
Pb* - measured burst pressure in the test
Projection processing must be used to utilize these methods for
spiral flaws. Three other different assessment methods for Q - spiral correction factor
spiral defects in linepipe were proposed by Mok et al., Fu and R - pipe radius
Jones, and Bai et al. separately. But these methods were
SMYS - specified minimum yield strength
originally developed for spiral corrosion and have not been
proved to be applicable for spiral weld defects on the old S r - ratio of applied load to flow stress
pipelines of PetroChina. In this paper’s work, several burst t - wall thickness of the pipe
tests of pipeline with spiral weld defects were carried out. W - width of the defect
Different projection assessment methods based on  - angle of the spiral defect
BS7910:2005 were conducted as well as the spiral corrosion
assessment methods proposed in previous studies. Predicted  flow - flow stress
results were compared with burst tests and analyzed in this  - dimensionless geometric parameter of the defect
paper. As a conclusion of comparison, the axial projection  - Poisson’s ratio
method based on BS7910:2005 was suggested to assess this
kind of spiral weld defects despite of its conservatism. INTRODUCTION
Some long distance transmission pipelines of PetroChina
KEY WORDS were constructed in the 1970s. Due to limited pipe
pipeline, spiral weld defect, burst test, engineering critical manufacturing technology at that time, some severe spiral weld
assessment (ECA) defects with large depth and large length existed on some

1 Copyright © 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 07/15/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


certain pipelines, which had been a great threat to pipeline The spiral weld defects were found to be lack of
integrity. A typical accident of spiral weld cracking was shown penetration and lack of fusion through excavation validation
in Figure 1, which took place in 2006 on Hui-Ning oil pipeline results. But there is no applicable ECA method for this kind of
in service since 1978[1]. How to inspect and assess these spiral defect up to now. To research for applicable assesssment
weld defects and make corresponding repair decisions became method, full scale burst tests of pipes with real spiral weld
an urgent issue. defects were conducted. These pipes were taken from an old
pipeline of PetroChina which had been inspected by tri-axis
MFL in-line inspection tool.
Two projection methods based on BS7910:2005 and
another three assessment methods for spiral corrosion were
summarized and applied to assess the defects of test pipes with
defect sizes reported by in-line inspection. Predicted burst
pressures of different methods were compared to test results to
find best practice to assess this kind of spiral weld defects.

ASSESSMENT METHODS FOR SPIRAL WELD


DEFECTS
BS7910:2005[3]
The British standard BS7910 is often used for assess flaws
in metallic structures, but for spiral weld defect the plane of the
flaw is not aligned with a plane of principal stress. Stress
intensity factor (SIF) and reference stress solutions are not
provided in BS7910:2005. Instead it is suggested to project the
Figure 1 - A leakage accident of spiral weld cracking flaw on to each of the three planes normal to the principal
stresses and to evaluate each of the three projected flaws. As an
Efforts were made to detect these defects through tri-axis example shown in Figure 3, the spiral flaw is projected
MFL in-line inspection. Through pull-through tests and trial respectively to the circumferential orientation and axial
application, it was proved that tri-axis MFL in-line inspection orientation. According to BS7910:2005, one of these
tool was able to detect and report sizes of spiral weld defects at projections will lead significantly high SIF and reference stress
a satisfying level. Figure 2 shows an example of tri-axial MFL than others and assessment is carried out on this projection. But
in-line inspection signals of spiral weld anomaly[2]. Estimated it is also reminded that special attention should be paid when
depth and length of the anomaly can be given through sizing the angle between the plane of the actual flaw and the principal
model on tri-axial signals. plane is greater than 20°. So in this paper’s work the
circumferential projection and the axial projection and further
calculations were carried out separately.

1-Resolved flaw 3-Principal axial stress


2-Actual flaw 4-Principal hoop stress
Figure 3 - Projection example for spiral flaw

In BS7910:2005 flaws are divided to 3 types: planar flaws,


Figure 2 - Example of tri-axial MFL in-line inspection non-planar flaws and shape imperfections. Generally planar
signals of spiral weld anomaly flaws result in fracture crack and non-planar flaws result in
plastic collapse. The spiral weld defects found on old pipelines
of PetroChina are mainly found to be due to lack of penetration

2 Copyright © 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 07/15/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


and lack of fusion. In this paper’s work the defects on test pipes
were not identified in detail because it is conservative to assess
non-planar flaws as planar ones.
For fracture assessment there are three levels of assessment
routes in BS7910:2005. Level 1 is the simplified assessment
method when the information of material properties is limited.
Level 2 is the normal assessment route, and level 3 is
appropriate for ductile tearing analysis. Level 1 assessment is a
conservative procedure compared to level 2 and 3. If the defect
can be accepted at level 1 assessment, there is no need to
employ high level assessment. In this paper’s work level 1A
and failure assessment diagram (FAD) was adopted for
conservative reasons.
As shown in Figure 4, level 1A FAD has two non-
dimensional variables. The vertical axis of is ratio of applied
conditions to material resistance to fracture, in terms of fracture
mechanics, corresponding to failure by fracture. The horizontal
axis is the ratio of the applied load to material flow stress, Figure 5 - Spiral correction factor Q
corresponding to failure by plastic collapse. The acceptable
region of level 1A FAD is bonded by two assessment lines: In Mok’s studies, the flow stress was set to 1.5SMYS
Kr  1 2  0.707 and S r  0.8 . So the safety factors and predicted results showed good agreement compared to
burst tests.
are 2 and 1.25 respectively. Method by Fu and Jones[6]
Fu and Jones made modifications on NG-18 equation and
introduced a spiral correction factor A( ,  ) together with
Folias factor M ( ) :
 d 
2 flow  t  1  
Pb  1  t  (2)
D  d 1 
 t A( ,  )  M ( ) 

Where  is the spiral angle of defects and
  4 12(1  2 )c / Rt is the dimensionless geometric
parameter of defects.
In their studies, an elastic finite element (FE) analysis was
Figure 4 - Level 1A FAD
conducted to determine mode I and mode II SIFs. An linear
regression of Folias factor was proposed for easy calculation as
After projecting flaws to the circumferential orientation or
Equation (3). The spiral correction factor A( ,  ) can be
to the axial orientation, the procedure for calculating K r and
obtained from Figure 6. Not as common projection methods,
Sr can be found in BS7910:2005. the total length of defect was used instead of projected length
Method by Mok et al. [4][5] because the FE analysis was based on the total length.
Mok et al. proposed a method to assess long external spiral M ( )  0.49  0.67 (for   1.5 ) (3)
corrosion:
2 flow  t  d
Pb  1  Q  (1)
D  t
Where Q is the spiral correction factor determined by
spiral angle  and ratio of defect width to wall thickness
W t . As shown in Figure 5, Q lies between 0.0 to 1.0. For
  60 or W t  32 , Q equals 1.0.

3 Copyright © 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 07/15/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


A  2 3 Ld (for L2 Dt  30 )
A  0.85 Ld (for L2 Dt  30 )
When Q is set to 1.0 for longitudinal defects
(   60 ), the criterion is quite similar to the original NG-18
equation.
In Bai et al.’s paper, compared to several published burst
tests, predicted results of this method were consistent with test
results.

BURST TESTS
To confirm actual burst pressure of pipeline and compare
with assessment methods, 7 groups of full scale burst tests of
pipes with real spiral weld defects were conducted at Langfang,
China in 2010.
The test pipes were taken from an old transmission
pipeline of PetroChina which had been inspected by tri-axis
MFL in-line inspection tool. The parameters of test pipes are
shown in Table 1. The spiral weld defects on them were real
defects formed in pipe manufacturing which were reported by
tri-axis MFL tool. Table 2 shows defect size parameters of the
test pipes. The relative defect depths (d/t) of the 7 defects were
close from 0.506 to 0.689 mm/mm. The lengths of defects were
all very long from 800mm to 6000mm. Some long defects were
defect cluster according to interaction rules.

Table 1 - Properties of the test pipes


Figure 6 - Spiral angle correction factor A( ,  ) Steel grade X52
Outer diameter 700mm
Wall thickness 8~9 mm
The dimensionless geometric parameter  was given in
Pipe length 12m approximately
the range of 0~8, so application of this method is limited for
Spiral angle 30°
long defects.
SMYS 358MPa
Method by Bai et al. [7]
Yield stress 381MPa
Bai et al. also made modifications on NG-18 equation and
Ultimate tensile strength 556MPa
combined the spiral corrosion factor Q from Mok et al. and
Fracture toughness 86.6MPa·m1/2
the Folias factor. The total length of defect was also used to
calculate Folias factor M in this method.
 A  Table 2 - Defect parameters
 1 Q  Test Wall thickness Defect depth Defect axial length
2 flow  t A0
Pb  1   (4) No. (mm) d/t(mm/mm) (mm)
D  1 A  1  1# 9 0.606 1900
 A0 M  2# 9 0.586 800
 
3# 8 0.515 5190
Where
4# 8 0.506 1490
2.51( L 2) 2 0.54( L 2) 4 5# 9 0.689 2017
M  1  6# 8 0.634 6000
Dt ( Dt ) 2
7# 9 0.573 1290
(for L2 Dt  50 )
L2 The burst test system mainly consists of hydraulic pressure
M  0.32  3.3 (for L2 Dt  50 ) unit, data acquiring and controlling unit and charge-coupled
Dt device (CCD) video camera as shown in Figure 7. The test pipe

4 Copyright © 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 07/15/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


was pressurized using water and monitored by video camera.
Strain and pressure history data of the test pipe were acquired
and recorded on the remote computer.

Data acquiring and controlling unit Remote monitor

Pressure control Pressure data Strain data CCD camera

Hydraulic
Test pipe
pressure unit
Figure 7 - Diagram of burst test setup Figure 8 - Cracking at the middle of weld - test 5#

The test results are summarized in Table 3 including burst


pressures and initial failure positions. It should be noticed that
for test 3# and 5#, the initial failure positions were at the
middle of weld and their burst pressures were significantly
higher than other 5 tests. For the other 5 tests, the initial failure
positions were on the fusion line and their burst pressures were
relatively low. Among them, 4 tests presented quite close burst
pressures from 5.3MPa to 5.9MPa. Photos of cracking at the
middle of weld and along the fusion line were shown in Figure
8 and 9 respectively for test 5# and 1#. It can be inferred that
the weld reinforcement will increase burst pressure when the
initial defect lies at the middle of the spiral weld, as the effect
of increasing wall thickness.
Figure 9 - Cracking along the fusion line - test 1#
Table 3 - Test results
Burst pressure Initial failure
Test No.
(MPa) position
ECA FOR SPIRAL WELD DEFECTS
1# 5.5 On the fusion line
Several ECA methods discussed above were applied to
2# 5.8 On the fusion line
predict burst pressure of the tests. Parameters of the pipe body
At the middle of material and the nominal wall thickness were used in
3# 9.6
weld calculation, i.e., the spiral weld defects were taken as defects
4# 5.9 On the fusion line on pipe body and weld reinforcement (remaining height) was
At the middle of not considered. It’s an approximate processing for easy use in
5# 9.4
weld practice.
6# 5.3 On the fusion line The ratio of actual burst pressure to predicted pressure
7# 7.2 On the fusion line
Pb* Pb was used to measure accuracies of different ECA
*
methods. For a “perfect” prediction, Pb Pb equals 1.0. For
“over predicted”, Pb* Pb is less than 1.0, and for “under
predicted”, Pb* Pb is more than 1.0. All the values of
Pb* Pb through different ECA methods are summarized in
Table 4.

5 Copyright © 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 07/15/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Table 4 - Actual burst pressure to predicted pressure
Pb* Pb
Test No. BS7910 Method by Mok Method by Bai
circumferential axial 1.5SMYS 1.1SMYS 1.5SMYS 1.1SMYS

1# 0.65 1.51 0.54 0.74 0.52 0.71


2# 0.65 1.46 0.57 0.77 0.54 0.74
3# 1.04 2.47 1.02 1.39 0.98 1.34
4# 0.63 1.46 0.62 0.85 0.60 0.82
5# 1.41 3.27 0.97 1.33 0.92 1.26
6# 0.76 1.80 0.60 0.82 0.57 0.78
7# 0.79 1.81 0.70 0.95 0.67 0.91
Average 0.84 1.97(1.61*) 0.72 0.98 0.69 0.94
Standard
0.28 0.67(0.18*) 0.20 0.27 0.19 0.26
deviation
*statistics excluding test 3# and 5#

Projection methods based on BS7910:2005


The spiral weld defects on test pipes were projected to the As shown in Table 4, burst pressures of 5 tests were over
circumferential and the axial orientation and then assessed predicted through this method. Only for test 3# and 5#, with
separately. These defects found to be due to a lack of initial failure position at the middle of weld, burst pressures
penetration and lack of fusion were mostly internal surface or were under predicted pressures. The average ratio Pb* Pb
embedded flaws. Tri-axis MFL in-line inspection tool can
report sizes of spiral weld defects but cannot classify their equals 0.84 with a standard deviation of 0.28.
position up to now. Generally internal surface flaws are more For the axial projection method, the spiral weld defects
critical than embedded and external surface flaws. For were projected to the axial orientation of the pipe. Solutions for
conservative considerations all these spiral weld defects were long internal surface flaw in cylinder oriented axially were
taken as internal surface flaws. used as shown in Figure 11.
For the circumferential projection method, the spiral weld
defects were projected to the circumferential orientation of the
pipe. Since the defects were relatively long, solutions for long
internal surface flaw in cylinder oriented circumferentially
were used as shown in Figure 10. For some very long defect,
the projected length was limited to not more than the
circumference of the pipe.

Figure 11 - Long internal surface flaw in cylinder


oriented axially

As shown in Table 4, burst pressures were under predicted


pressures for all 7 tests through this method. Although the
*
ratios Pb Pb were relatively large (average 1.97 with a
standard deviation of 0.67) indicating conservatism of this
method. Especially for test 3# and 5#, the ratios Pb* Pb were
2.47 and 3.27 respectively. For other 5 tests, the ratios were
somehow close from 1.46 to 1.81. If statistics are only made on
the 5 tests, an average ratio Pb* Pb of 1.61 with a standard
Figure 10 - Long internal surface flaw in cylinder oriented deviation of 0.18 is readout. Under this condition we will have
circumferentially

6 Copyright © 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 07/15/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


an average safety factor of 1.61 and a minimum safety factor For three methods for spiral corrosion, the method by Fu
1.46 for the 5 tests. and Jones was not employed in this paper because defect length
Method by Mok et al. here exceeded its application range. The method by Mok et al.
To utilize this method, width parameter W of defects is and by Bai et al. presented quite close burst pressures
required. However the defect width was not reported by tri-axis compared with each other. But they both over predicted the
MFL in-line inspection tool. Excavation validation showed burst pressures for most tests even if the flow stress was
most spiral weld defects were approximately 2~3mm in width. decreased from 1.5SMYS to 1.1SMYS . This probably
In this paper W was set to 4mm. Actually when spiral angle because these two methods were based on tests of external
equals to 30°, Q varies very slightly when W changes in a surface spiral corrosion defects, which were quite different
from the spiral weld defects in this paper’s tests.
certain range.
Among these methods above, only the axial projection
As shown in Table 4, when flow stress was set to
method based on BS7910:2005 is appropriate to assess the
1.5SMYS , burst pressure was over predicted for 6 tests with spiral weld defects on the test pipes in spite of its relatively
an average ratio Pb* Pb of 0.72 and a standard deviation of high conservatism.
0.20. When flow stress was decreased to 1.1SMYS , still 5
*
CONCLUSIONS
tests were over predicted with an average ratio Pb Pb of Studies were carried out to solve the issue of assessing
0.98 and a standard deviation of 0.27. spiral weld defects on some old pipelines of PetroChina.
Method by Fu and Jones Several possible ECA methods for the spiral weld defects were
All the spiral weld defects on test pipes in this paper were summarized, including two projection methods based on
very long (   25.5 ~ 203.2 ), which had greatly exceeded BS7910:2005 and other three methods for assessing spiral
the range of  provided in Fu and Jones’s paper. So this corrosion.
Seven groups of full scale burst tests of pipes were
method was not employed for assess these spiral weld defects
conducted. These test pipes with real spiral weld defects were
in this paper.
taken from an old pipeline which had been inspected by tri-axis
Method by Bai et al.
MFL in-line inspection tool. It is noticed that when the initial
The same assumption of W was made with the method failure positions were at the middle of weld, burst pressures
by Mok et al., i.e., the same value of Q was used. The would be significantly higher than other tests due to effect of
predicted results by this method were also quite close to results weld reinforcement.
by Mok’s method. When flow stress was set to 1.5SMYS , all Different ECA methods were employed with defect size
parameters reported by in-line inspection tool. The predicted
the 7 tests were over predicted with an average ratio Pb* Pb results of different methods were compared with test results
of 0.69 and a standard deviation of 0.19. When flow stress was and analyzed.
decreased to 1.1SMYS , also 5 tests were over predicted with Methods for assessing spiral corrosion were found to be
an average ratio Pb* Pb of 0.94 and a standard deviation of not appropriate to assess the spiral weld defects on the test
pipes probably due to different defect types and positions. The
0.26. circumferential projection method based on BS7910:2005 also
Summary over predicted burst pressures for 5 tests. Only the axial
For two projection methods based on BS7910:2005, burst projection method presented conservative burst pressures for
pressures of 5 tests were over predicted through the all 7 tests.
circumferential projection method. Only the axial projection It is concluded that the axial projection method based on
method presented conservative predicted burst pressures for all BS7910:2005 was suggested to assess this kind of spiral weld
tests. The ratios of actual burst pressure to predicted pressure defects on the old pipelines of PetroChina despite of its some
Pb* Pb varied from 1.46 to 3.27 with an average value of 1.97 kind of conservatism. It’s valuable to make repair decisions
and a standard deviation of 0.67, indicating relatively high through this method from the large amount of spiral weld
conservatism of this method. But if test 3# and 5#, affected by defects on the old pipelines.
weld reinforcement, are not count, the average ratio of Pb* Pb ACKNOWLEDGMENTS
will be 1.61 with a standard deviation of 0.18. Under this The authors would like to thank PetroChina Pipeline
condition the accuracy and consistency of the axial projection Company for permission to publish this paper. The authors also
method seem to be good enough to assess the spiral weld would like to thank PetroChina Pipeline Company staff’s
defects. However, the population size is not big enough to infer valuable help in gathering in-line inspection data, collecting
a realistic safety factor and a standard deviation. More tests are test pipes and the burst tests.
needed for further investigation.

7 Copyright © 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 07/15/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use


REFERENCES
[1] Qingshan Feng, Yi-Han Lin, Fuxiang Wang, and Bin Li,
Preliminary Failure Assessment for Spiral Welded Defects of
Pipeline, proc. IPC2008, 7th Int. Pipeline Conf., Sept, Oct,
2008, Calgary, Canada, ASME, IPC 2008-64059.
[2] Fuxiang Wang, Qingshan Feng and Lijian Zhou,
Application Progress of Tri-Axial MFL Sensors Technology,
ASCE Conf. Proc. ICPTT 2011: Sustainable Solutions for
Water, Sewer, Gas, and Oil Pipelines, Proceedings of the
International Conference on Pipelines and Trenchless
Technology; pp. 981-989.
[3] British Standard BS7910: Guide on methods for assessing
the acceptability of flaws in metallic structures, 2005.
[4] Bjornoy, O. H. and Marley, M. J. (2001). Assessment of
Corroded Pipelines, Past, Present and Future. Stavanger,
Norway: ISOPE 2001, Proceedings of the 11th International
Offshore and Polar Engineering Conference. June 17-22. Vol
II. 93-101.
[5] D.H.B. Mok, R.J. Pick, A.G. Glover, R. Hoff, Bursting of
line pipe with long external corrosion, International Journal of
Pressure Vessels and Piping, Volume 46, Issue 2, 1991, Pages
195-216.
[6] Fu, B., and Jones, Failure of Spiral Corrosion in Linepipe.
The 13th International Conference on Offshore Mechanics and
Arctic Engineering, Houston, TX, USA, 02/27-03/03/94; pp. 1-
8. 1994.
[7] Bai Y., Xu T. and Bea, R. Reliability-Based Design and
Requalification Criteria for Longitudinally Corroded Pipelines.
Proceedings of the Seventh (1997) International Offshore and
Polar Engineering Conference, Honolulu, USA, May 25-30,
1997.

8 Copyright © 2012 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 07/15/2018 Terms of Use: http://www.asme.org/about-asme/terms-of-use

You might also like