You are on page 1of 22

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

IPC2012-90597

INVESTIGATION INTO THE FAILURE OF A 40 DIAMETER CRUDE OIL PIPELINE


Mohamed Dafea mohamed.dafea@gl-group.com GL Noble Denton Holywell Park, New Ashby Road, Loughborough, LE11 3GR, UK Tel: +44 1509 28 2292, Fax: +44 1509 28 3119 Phil Hopkins p.hopkins@penspen.com Roland Palmer-Jones r.palmer-jones@penspen.com

Penspen Ltd. Unit 7-8, Terrace Level, St. Peters Wharf, Newcastle-upon-Tyne, NE6 1TZ, UK Tel: +44 1912 38 2200, Fax: +44 1912 75 9786 Patrick de Bourayne pdebourayne@trapil.com Trapil 7-9 rue des Frres Morane, 75738 Paris Cedex 15, France ABSTRACT In August 2009, there was a 2.5 m long rupture in the longitudinal seam weld of a crude oil pipeline in France. The failure caused a spillage of approximately 2000 cubic metres in a protected area. This rupture caused the authorities to withdraw the permit to operate a 260 km long section of this pipeline. The pipeline had a similar failure in August 1980. The 1980 failure was attributed to a fatigue crack initiating at the inner side of the longitudinal weld. There was evidence of roof topping along this weld. Penspen Ltd., UK were contracted by the pipeline operator Societe du Pipeline Sud Europeen (SPSE) to carry out an independent investigation into the cause of the 2009 failure, review and confirm the actions needed for safe short term operation to allow internal inspection, and determine a safe future life for the pipeline. The failed pipe specimen was not immediately available for inspection. An initial analysis indicated that purely an analytical evaluation would not provide conclusive results, due to variability in material properties, geometry and loading. It was decided that a better understanding of the behaviour of defects in the pipe, and the fatigue performance, was required. A detailed laboratory programme of burst and fatigue testing on a section of line-pipe was recommended. The tests Lionel Blin lionel.blin@SPSE.FR Societe du Pipeline Sud Europeen La fenouillire Route d'Arles BP 14 13771 Fos sur Mer Cedex, France were carried out on a combination of defect-free1 ring specimens, and ring specimens with initial Electric Discharge Machined (EDM) slits to represent crack-like defects. These tests showed: 1. BURST TESTS: The failure pressures of the burst tests could be predicted using a recognised industry model. Roof topping, laminations and inclusions, and toughness variations were found to have no noticeable effect on the defect size at failure. 2. FATIGUE TESTS: The fatigue test results showed that defect-free rings had a fatigue life one order of magnitude longer than those containing an EDM slit, and that the fatigue life of a defect-free ring could be predicted using a standard SN method. In addition, it was found that crack growth was conservatively predicted using standard fatigue fracture mechanics. The fatigue test results showed that the 1980 and the 2009 failures were caused by a combination of cyclic pressure loading, roof topping, and a pre-existing weld defect (probably present when the pipeline went into service in 1972). This paper provides an overview of the failure investigation, with a discussion of the ring testing, supporting tests, and fracture surface inspection.

1 Defect-free means that any deviation from sound pipe or weld material is within workmanship limits, References [1] and [2].

Copyright 2012 by ASME

INTRODUCTION Societe du Pipeline Sud Europeen (SPSE) owns and operates a 40 diameter pipeline, approximately 700 km long, which delivers crude oil to several refineries in France. The pipeline was commissioned in 1972 [3], and is constructed of API 5L X60 [2] line pipe, with wall thickness varying between 8.74 mm and 12.7 mm. The pipeline is pressure cycled due to variations in the demand for the oil.
1.1. 1980 Failure

1.

In August 1980, the pipeline ruptured at a longitudinal weld, and evidence of inclusions and laminations was present at the fracture surface of the failed section of pipe. A failure investigation study [3] attributed the failure to a fatigue crack initiating at the inner side of the weld, combined with a geometry defect known as roof topping, see Figure 1. Roof topping is caused by failure to crimp the plate edges sufficiently before the U and O pipe forming process, leaving a peak along the seam weld. Roof topping may decrease the fatigue life of a pipeline by concentrating the local stresses at the seam weld. A failure investigation in 1980 focused on the following: 1. theoretical modelling of the stress concentration caused by roof topping; 2. defect depth to cause failure; and, 3. fatigue crack growth during service. Operational limits were set for the maximum allowable roof topping height, and maximum defect depth, based on this work. SPSE carried out numerous repairs, between 1980 and 2009, taking into account the 1980 failure investigation results and operational experience.
1.2. 2009 Failure

1. S-N analysis: an empirical approach based on fatigue tests. This has been used to investigate the cause of the 1980 and the 2009 failures2. 2. Analytical models based on fracture mechanics: this has been used to establish a conservative approach for determining the remaining fatigue life of the pipeline. The approach used the Battelle flow stress dependent equation [5]), to determine defect limits, and BSI 7910:2005 [6] to model seam weld fatigue crack growth from an initial crack size to the critical size, where there is a geometry defect increasing the local stresses (e.g. roof topping and weld misalignment). 2. D EDM Nlowerbound : Nmean : Nupperbound : P SCF t : : : : : NOMENCLATURE : : Pipe diameter (mm). Electric Discharge Machined Lower bound allowable number of cycles for fatigue design from an S-N curve, representing a 2.3% probability of failure. Allowable number of cycles to reach the mean S-N curve, representing a 50% probability of failure. Upper bound number of cycles from an S-N curve, representing a 97.73% probability of failure. Pressure range (bar). Stress concentration factor due to geometry defects e.g. roof topping, weld misalignment. Pipe wall thickness (mm). Hoop stress range resulting from internal pressure (MPa). Bending stress due to due to geometry defects e.g. roof topping, weld misalignment (MPa). Total stress range taking into account stress elevation due to geometry defects e.g. roof topping, weld misalignment (MPa).

In August 2009, another longitudinal weld failure occurred (see Figure 3 [4]), and evidence of inclusions and laminations was again present at or around the fracture surface of the failed weld. Penspen Ltd. UK, was contracted by SPSE to investigate the cause of failure and estimate the remaining fatigue life of the pipeline. The failed pipe section was not available for inspection or testing. Initial analytical work indicated that the failure may have been related to the presence of a crack-like seam weld defect in 1972, when the pipeline went into operation, or it might have been caused by a combination of cyclic loading and roof topping. It was decided to conduct a programme of small scale material tests and ring testing (Figure 2) to quantify the burst and fatigue performance of the pipeline. This testing was carried out by GL Noble Denton (GLND), UK. The ring tests, of line pipe cut from the pipeline, included 11 static (burst) tests and 13 fatigue tests. Following completion of the ring tests, Penspen used a two-step approach to analyse the test results and explain the 2009 failure:

3.

SMALL SCALE MATERIAL TESTS A series of small scale material tests was carried out on 3 pipe samples, obtained from the pipeline adjacent to the 2009 failure, to ensure a complete record of material properties was available. The following tests were carried out: 1. Longitudinal and transverse tensile testing of the parent pipe material. 2. Macro examination and hardness testing of the seam weld.
2 The analysis was based on both: the 13 fatigue ring tests carried out between 2010 and 2011, and the small scale fatigue tests carried out in 1980.

Copyright 2012 by ASME

3. Chemical composition of the parent pipe material. 4. Crack Tip Opening Displacement (CTOD) fracture toughness testing of the parent pipe material. 5. Sulphur print testing of parent pipe material. 6. Charpy impact testing of the parent pipe material and the longitudinal pipe seam weld. Small scale testing results indicated the following: Sample 1 and 2: The pipe meets the API 5L specifications in terms of tensile strength, Charpy impact toughness, hardness, and chemical composition, for X60 pipe material. Sample 3: o The pipe meets the API 5L specifications in terms of tensile strength, seam weld Charpy impact toughness, and seam weld hardness, for X60 pipe material. o Charpy tests of the parent material failed to meet the current minimum impact energy requirement of API 5L3 and recorded low percentage shear areas (12% and 15%), which can indicate low toughness. 4. RING TESTS A ring tension test rig is shown in Figure 2. A short ring of steel (approximately 76 mm in width) is cut from a pipeline and placed inside the test rig. A ring tension test rig can also be pressure cycled to investigate the effects of fatigue loading. The ring sample is then hydraulically expanded until the sample fails. Notches can be machined into the sample and a fatigue crack can be grown from the edge of the notch by repeated pressure cycling in the rig, to investigate the effects of such cracks.
4.1. Static Tests

to understand the failure behaviour; i.e., was the failure predominantly ductile or brittle.
4.2. Fatigue Tests

13 fatigue ring tension tests were also carried out. These tests consisted of two sets: 11 fatigue tests with initial EDM Slits. 2 fatigue tests without initial EDM slits. A summary of the fatigue test results is given in Table 3. EDM slits were chosen as it is very difficult to grow fatigue cracks to a prescribed depth and precise location in a seam weld with roof topping. The EDM slits behave similar to fatigue pre-cracks in this material (see Figure 14 and Figure 15 later). The key objective of the fatigue tests was to understand the development and growth of fatigue cracks at the longitudinal seam weld toe. Results of the above testing programme were analysed in order to: investigate the cause of previous failures; and, estimate the pipeline remaining fatigue life. A summary of this analysis is explained in the sections below. FAILURE INVESTIGATION The 1980 and 2009 failures were attributed to fatigue cracks initiating at the inner surface of the pipe, at the toe of the seam weld, where there were geometry defects known as roof topping. Fatigue is the progressive damage caused to a material when subjected to cyclic loading. This involves the formation and growth of a crack. Under repeated cycles, the crack may propagate to such an extent that total fracture of the material may occur below its maximum strength properties. Typical sources of cyclic loading in pipelines are: Cyclic loading during construction (e.g. transportation of line pipes can cause railroad cracking). Cyclic loading during service (e.g. internal pressure cycling). Fatigue is not a major cause of failures in oil and gas pipelines, but where they do occur, they are usually associated with longitudinal seam weld defects, or damage such as dents.
5.1. S-N Assessment Procedure

5.

11 static ring tension tests were carried out. These tests consisted of three sets: 1 static test on a ring on the as-received pipe (i.e. it did not contain any machined defect). 6 static tests with EDM slits4. 4 static tests with EDM slits and fatigue cracks at the base of the slit. A summary of the static test results is shown in Table 1 and Table 2. The key objective of these static tests was to be able to predict critical defect sizes defects that would cause failure at the pipelines operating pressure. A secondary objective was

The limit to the number and amplitude of stress cycles which a defect-free material can survive is generally plotted on an S-N curve. The procedures are based on the quantitative relationship between the fatigue strength and number of cycles corresponding to a specific probability of failure.
5.1.1 S-N Curve

With an average of 17 J versus minimum required of 30 J. 4 An EDM slit has a width of 0.15 mm, and resembles a sharp defect such as a planar weld defect.

The procedures described in BS 7608:1993 [7] have been used to assess the results of the fatigue tests and operational

Copyright 2012 by ASME

fatigue loading due to internal pressure cycling against S-N curve limits at a reference stress range, ref5, see Figure 4: 1. The lower bound allowable number of cycles for design (Nlowerbound), representing a 2.3% Probability of Failure (POF). 2. The allowable number of cycles to reach the mean curve (Nmean), representing a 50% POF. 3. The upper bound number of cycles (Nupperbound), representing a 97.7% POF. The weld classification chosen to represent a double-sided full penetration submerged-arc weld is Class E [7].
5.1.2 Reference Stress Range

5.3. S-N Assessment of Rings without Initial EDM Slits

Two defect free rings were fatigue tested and assessed against BS 7608 S-N curves. Of these, only one ring had roof topping. The other ring had no roof topping; however, the ring was oval and the weld was misaligned. These rings have fatigue lives predicted by the S-N curve5, see Figure 6. Further analysis of the fracture surface of one of these rings indicated that the ring experienced a mixture of brittle and ductile fracture just after the crack had grown through the middle third of the remaining wall (an area of banded inclusions and laminations).
5.4. S-N Assessment of the 1981 Fatigue Tests

A reference stress range of 317 MPa has been selected to represent the cyclic stress range caused by the pressure variations in the pipeline6, and the equivalent number of cycles for various pressure ranges has been calculated using Miners 7 rule [8]. This stress range is an approximate representation of the pipeline minimum and maximum operating pressures (10 bar and 29 bar), and a roof topping height of 4.5 mm. The method used to calculate the equivalent number of cycles at this reference stress range for any ring is based on the conversion of a pressure range, P, into a stress range, , taking into account the stress concentration factor resulting from geometry defects5 (e.g. roof topping and weld misalignment).
5.2. S-N Assessment of Rings with EDM Slits

11 fatigue tests were carried out on rings with EDM slits with depths between 1.3 mm and 4.3 mm8. Of these 11 rings, 2 had laminations, and similar slit depths at the start of the fatigue tests; however, their fatigue lives were dissimilar: one ring had a fatigue life 3 times longer than the other ring. The test results suggests that laminations do not shorten the fatigue life, but the effect of laminations on fatigue life is difficult to predict, as it depends on its location relative to the pre-existing defect, as well as its shape and continuity. Figure 5 shows the results of the S-N assessment of all fatigue tests on rings with EDM slits (EDM slit depth given in brackets after the ring number). All rings with EDM slits had a fatigue life shorter than that of the lower bound predicted by BS 7608. This is expected as the S-N curves do not account for the presence of a defect such as an EDM slit.

8 small scale fatigue tests (four point bending tests) had been carried out by Centre des Matriaux de l'Ecole des Mines in 1981 following the 1980 failure [9] on 2 samples (designated F and J) of the pipeline. Sample F had a large roof topping, offset of the external weld to the internal weld, and also contained micro cracks with depths between 15 m and 100 m (within weld workmanship limits). Sample J did not have any roof topping. The samples were cycled between 100 MPa and 500 MPa. Specimens F1, F8 and J1 were halted prior to failure because there was a crack approximately 3 mm in length. Specimen F7 was tested at a different stress range which was not provided in the test report. Figure 7 shows the comparison of these 1981 test results with the Class E S-N curve5. The number of cycles was corrected so that the data could be plotted at the reference stress range9. All samples, except F1, had a fatigue life higher than the upper bound limit given in BS 7608, so the S-N curve provides a conservative prediction of the fatigue lives of these defectfree specimens, and are similar to the defect-free rings in Figure 6.
5.5. S-N Fatigue Assessment of Previous Failures

5 Stress calculations take into account the elevation in stress resulting from roof topping and weld misalignments. These were calculated using BS 7910 for each ring, and validated using finite element (FE) stress analysis. 6 From most recent operating regime. 7 Assumes linear damage i.e. the combined effect of all cycles will cause the structure to fail when the total damage = 1. 8 All EDM slits were positioned at the internal surface of the weld toe, on the misaligned side of the weld to match previous pipeline failures.

Pipeline pressure history data was available for the period between 1994 and 2008. This data was analysed10, taking into account the stress elevation caused by roof topping, and an estimated number of completed pressure cycles at the reference stress range was calculated for the 1980 and the 2009 failures. This data was compared with the BS 7608 Class E S-N curve, see Figure 8. Figure 8 indicates that the fatigue loading applied up to the time of the 1980 and 2009 failures was not sufficient to cause failure if the welds contained only workmanship anomalies (i.e. no pre-existing defect was in the weld), despite the roof topping: It has been estimated that 40 years of service would have been required for the section of pipe with 7.5 mm
9 Note that variations in roof topping across the samples were not accounted for in this analysis, as they were tested directly in bending. 10 Pressure cycles were counted using the reservoir cycle counting method, see Reference [7].

Copyright 2012 by ASME

of roof topping that failed in 1980 to reach a 50% probability of failure. In fact it failed after just 8 years. For the section of pipe that failed in 2009, which had 4.5 mm roof topping, 104 years of service would have been required, rather than 36 years. Figure 9 shows that defect-free rings have a fatigue life approximately 1 order of magnitude greater than rings with defects (EDM slits). The 1980 and 2009 failures have fatigue lives similar to those of rings with EDM slits. This suggests that there were pre-existing defects (e.g. weld defects) at these failure locations. It is noted that: 1. The cycles to failure for the 2009 failure (a lamination present, and possibly, a pre-existing defect) was 28,623 cycles11. 2. The cycles to failure for one ring (initial EDM of 2.7 mm and a lamination) was 29,256 cycles11.
5.6. S-N Assessment of Previous Failures

Conservatively predict the defect size to cause pipeline failure. Conservatively predict the fatigue loading required to grow a defect from an initial known or assumed size to a size to cause failure (the critical size).
6.1.1 Model to Determine Critical Defect Size

The recorded failure pressures and defect depths prior to failure of the ring specimens were compared with the values predicted using a recognised model from the pipeline industry: the Battelle flow stress dependent equation [5]. This is to establish a relationship between the depth of a defect in the toe of the weld, and the failure stress of the pipe sample. The Battelle equation was developed in the USA in the 1960s by the Battelle Memorial Institute to predict the failure behaviour of the line pipe containing defects. Battelle carried out a large number of full scale tests (over 300) and developed a semiempirical model. This model is now widely-used to predict the failure pressure of a part wall defect in thin-walled (low constraint), ductile line pipe steel13.
6.1.2 Model to Determine Fatigue Life

The equivalent number of years of operation at the reference stress range (317 MPa) for each ring was calculated from testing based on the pipeline pressure history between 1994 and 2008, see Figure 10. This Figure shows when each specimen would be expected to fail, if it had been in the pipeline. Figure 10 shows: 1. Defect-free rings have a fatigue life one order of magnitude higher than rings with EDM slits. 2. The reduction in fatigue life due to the bending stress introduced by roof topping is at least one order of magnitude12. 3. The fatigue life of the two operational failures (1980 and 2009) coincide with ring tests containing EDM slots. The results of the fatigue testing indicate that there were pre-existing weld toe defects present at the locations of the 1980 and 2009 failures. These defects were probably present when the pipeline went into service. This suggests that the 1980 and the 2009 failures were not caused by roof topping alone: a combination of pressure cycling, roof topping, and a preexisting defect caused these failures. 6. REMAINING FATIGUE LIFE CALCULATIONS

6 rings were monitored during fatigue testing and the number of cycles to failure (excluding number of cycles for crack initiation at the slit) was compared with the values predicted by BSI 7910:2005 [6]. This was to establish a conservative method for prediction of remaining fatigue life of an existing defect in the pipeline.
6.2. Burst Assessment of Static Test Results on Rings with EDM Slits

6.1. Introduction

Conservative engineering models are needed, in order to establish a conservative methodology to evaluate the remaining fatigue life of the pipeline. These engineering models must:

7 rings with EDM slits, and 1 ring without an EDM slit, were tested. Of these, 1 ring had a mid-wall lamination and an EDM slit, and was tested to evaluate the effect of laminations on the failure pressure. All rings had roof topping, see Table 1. All rings showed a predominantly ductile failure. However, a small area of brittle fracture is evident near the outer surface of each ring. In addition, inclusions present in the steel are evident in the mid-thickness. A typical example is given in Figure 11 for the fracture surface of a ring, with the relevant micrographs of the surface in Figure 12. The results of burst tests on EDM slits are illustrated in Figure 13. The results on EDM slits match the failure pressures predicted using the Battelle equation. These ring tests show that the critical defect depth for the pipeline at a pressure of 40 bar (resulting in a hoop stress of 233 MPa) is approximately 5 mm (see Figure 13).

At a reference stress range of 317 MPa. The 1980s fatigue tests were four points bending tests. These tests will not capture the effect of stress elevation due to roof topping.
12

11

13

Note that all defects initiated in a ductile manner.

Copyright 2012 by ASME

6.3. Burst Assessment of Static Test Results on Rings with EDM Slits and Fatigue Pre-Cracks

2. Convert pressure ranges into hoop stress ranges as [11]: 3. Calculate SCFs for each ring, resulting from roof topping at the minimum and maximum pressures of a pressure range using BS 7910. 4. Calculate the bending stress range using BS 7910 (Annex D) as:
=( * (SCF -1))@min. pressure ( * (SCF -1))@max. pressure

3 rings with EDM slits and fatigue pre-cracks14 were tested. The fracture surfaces of the tests with fatigue cracks grown from the EDM slit are similar to those from tests on EDM slits only, i.e. predominately ductile with a small brittle area near the outer surface. The results of burst tests on EDM slits and fatigue cracks are illustrated in Figure 14. These results show that the failure pressures of rings with EDM slits and fatigue cracks in this line pipe material are both reasonably predicted by the Battelle model.
6.4. Burst Assessment of Fatigue Tests

The ring tests used to assess the effect of fatigue were investigated to give more burst test results: this was achieved by estimating the fatigue crack depth at failure. The results are shown in Figure 15, and add more predictions for cracks in the weld. The above results show that the failure pressures of rings with EDM slits or fatigue cracks are again reasonably predicted by the Battelle model. The burst test results show that: The failure pressures of the ring tests were reasonably predicted using a recognised industry failure model. Roof topping and laminations have no noticeable effect on the burst pressure, or critical defect size for failure. The pipe failure is predominately ductile, and the failure pressure depends on the tensile strength of the steel, rather than the toughness.
6.5. Analytical Modelling of Fatigue Test Results: Future Operation

5. Calculate the number of cycles required to grow a crack from an initial depth to a final depth at the loading conditions above (hoop and bending stress ranges) using BSI 791015. A comparison between the numbers of cycles recorded during the ring fatigue tests and the fatigue life predictions made using this assessment approach is shown in Table 4. The results of this comparison show that the fatigue life of a section of this pipe with roof topping and a crack like seam weld defect is conservatively predicted using BSI 7910 (with a minimum safety factor of 3.6, see Table 4). This methodology can therefore be used to predict the fatigue life of known defects in the pipeline16, or defects that could be present, but are below the detection threshold of any in line inspection system used in the future. 7. CONCLUSIONS 1. Static ring test results show: a. The failure pressures of the ring tests were reasonably predicted using a recognised industry model. b. Roof topping and laminations have no noticeable effect on the burst pressure, or critical defect size for failure. c. The pipe failure is predominately ductile, and the failure pressure depends on the tensile strength of the steel. 2. Fatigue ring test results show: a. The fatigue life of the pipeline can be conservatively predicted using a recognised industry model. b. The reduction in fatigue life due to the bending stress introduced by roof topping is at least one order of magnitude. c. Rings with pre-existing defects have a fatigue life one order of magnitude less than defect-free rings. 3. Causes of pipeline failures: This study has shown that the 1980 and the 2009 failures of the SPSE

Fatigue testing of each ring was carried out at different pressure ranges in order to simulate operational conditions, see Table 3. A retrospective study of fatigue crack growth on 6 rings has been undertaken in accordance with BSI 7910, using the TWI CrackWise software [10]. These rings were monitored during testing, by regular visual checks, and the number of cycles to initiate a crack evident at the edge of the ring was recorded. Consequently, it was possible to exclude the cycles for crack initiation from the fatigue assessment, and only assess the growth of the fatigue crack. The following assessment procedure was followed: 1. Calculate the critical crack depth at the maximum operating pressure using the Battelle flow stress dependent equation.

14 Fatigue cycles were applied to these rings to generate initial fatigue precracks. The rings were then pressurised to failure, and the average depth of the fatigue pre-crack was measured from the fracture surface.

15 Based on the upper bound values of BSI 7910 for Paris fatigue crack growth law (constants: m = 3, and A = 2.3 X 10-12). 16 The information on critical defect sizes will also be useful when planning future inspections and evaluating the crack detection thresholds of different tools.

Copyright 2012 by ASME

pipeline in France were a combination of pressure cycling, roof topping, and a pre-existing defect. 4. Future operation: A conservative methodology for predicting the remaining fatigue life of the pipeline has been developed and validated by testing. ACKNOWLEDGMENTS The authors would like to acknowledge the support of SPSE, the SPSE shareholders (Total S.A., ExxonMobil Corporation, Socit de Participations dans l'Industrie et le Transport du Ptrole, BP France, Socit des Ptroles Shell, BASF SE, CococoPhillips Germany GmbH), and in particular, Wilson SantaMaria, of ExxonMobil, UK for valuable support and discussions throughout the course of this work and permission to publish. The diligent testing carried out by GLND, UK is also gratefully acknowledged. REFERENCES [1] Cosham, A., Hopkins, A., 2004, An Overview of the Pipeline Defect Assessment Manual (PDAM), 4th International Pipeline Technology Conference, Oostende, Belgium. [2] Anon., 2007, Specification for Line Pipe, ANSI/API Specification, API 5L, 44th Edition. [3] Hourlier, F., Pineau, A., 1981, Bursting of a Bergrohr Tube of The S.P.S.E. Pipeline, Final Report, Ministere de LIndustrie, Ecole National Superieure des Mines de Paris, Centre des Materiaux. [4] Anon.; 2010, Troncon Fissure No43440A Reception (5 Planches), Appendices of the 2009 Failure Investigation Report, Report No.: 4823115-001-1 PC. [5] Kiefner, J.F., Maxey, W.A., Eiber, R.J., and Duffy, A.R., 1973, The Failure Stress Levels of Flaws in Pressurised Cylinders, ASTM STP 536, American Society for Testing and Materials, Philadelphia, pp. 461-481. [6] Anon., 2005, Guide To Methods For Assessing The Acceptability Of Flaws In Metallic Structures, BSI 7910: 2005, Incorporating Amendment 1 (2007), British Standards Institution, London, UK. [7] Anon., 1995, Code of Practice for the Fatigue Design and Assessment Of Steel Structures, Incorporating Amendment No. 1, BS 7608:1993, British Standard. [8] Miner, M. A., 1945, Cumulative Damage in Fatigue, J. Appl. Mech., Vol. 12, Trans. ASME, Vol. 67, pp. A159A164. [9] Anon., 1980, Inspection and Tests performed on a Welded Pipe Burst in Operation, Institut de Soudure Test Report, 62 920. [10] Anon., 2009, Crackwise Software; Compiled by The Welding Institute, VERSION 4.1. [11] Anon., 2010, Gas Transmission And Distribution Systems, ASME Code For Pressure Piping B31, ASME B31.8-2010, Revision Of ASME B31.8-2007, The American Society Of Mechanical Engineers.

Copyright 2012 by ASME

ANNEX A: TABLES Maximum Roof Topping Height (mm) 3.0 3.1 3.5 3.6 3.8 4.2 4.1 Final Crack Average Depth at Failure (mm) 0.00 1.05 1.85 2.85 3.65 5.35 3.65

Ring No. 1 2 3 4 5 6 21

Measured EDM Slit Depth (mm) N/A 1.05 1.85 2.85 3.65 5.35 3.65

Failure Pressure (barg) 102.817 95.0 76.7 61.2 54.9 43.2 64.8

Comment No EDM EDM Slit Only EDM Slit Only EDM Slit Only EDM Slit Only EDM Slit Only EDM Slit Only

Table 1: Summary of Static Test Results on Rings with an EDM Slit Only.

Ring No.

Maximum Roof Topping Height (mm) 4.3 4.4 4.5 4.8

Fatigue Pre-Crack Initiation Measured EDM Slit Depth (mm) 0.85 0.95 10 to 40 6000 16800 8500 5000 3.08 2.1420 4.73 4.79 47.7 53.2 1.65 2.65 21 to 29 10 to 29 10 to 29 Pressure Range, bar Number of Cycles Fatigue PreCrack Average Depth (mm)18 0.0019 3.40 EDM Slit + Fatigue PreCrack Average Depth (mm) 0.85 4.35 Failure Pressure, barg

7 8 9 14

10 to 40 10 to 29

18000 28000

98.3 54.0

Table 2: Summary of Static Test Results on Rings with an EDM Slit and a Fatigue Pre-Crack.

17 18

Note this ring did not fracture but coned. The average depth of the fatigue pre-crack was measured from the fracture surface of the ring, after the test. 19 No measurable crack initiated after the ring was fatigue cycled before the static test. 20 Further fatigue cycling (2,000 cycles at a pressure range of 10 to 29 bar, and 4,000 cycles at a pressure range of 21 to 29 bar) was applied, but final fatigue precrack depth was not measured prior to the static test.

Copyright 2012 by ASME

Ring No. 10

Roof Topping Height (mm) 4.3

Measured EDM Slit (mm) 2.70

Pressure Range (bar) 10 to 29 18 to 32 10 to 29

Total Number of Cycles 3500 12024 1200 683 10305 6300 12, 460 12310 14, 000 1, 565 16753 7, 000 14000 668 164,493 5500 20, 000 11, 173 2000 7, 513 4376 4, 000 230,000 10, 165

11 12 13

4.4 4.4 4.1

3.65 10 to 40 2.55 2.60 18 to 32 10 to 29 10 to 29 10 to 29

15 17 18

4.6 4.4 4.4

2.00 17 to 29 4.25 1.35 17 to 29 10 to 29 10 to 29 10 to 29

19 20

4.4 4.5

1.35 17 to 29 N/A 15 to 40 10 to 29

22

5.2

2.70

18 to 32 15 to 35 10 to 29

23

5.0

2.80 15 to 35 10 to 29

24

3.8 N/A21

2.80 21 to 29 15 to 40 N/A 10 to 80
Table 3: Summary of Fatigue Test Results.

25

21

Ring had no roof topping; however, the ring was oval and the weld was misaligned.

Copyright 2012 by ASME

Ring No.

10 12 13

15

18 24

EDM Slit + PreCrack Depth22 (mm) 3.2 3.05 3.10 5.20 2.50 4.09 5.26 5.26 5.26 1.85 2.88 3.30 3.81

Crack Depth after Fatigue Growth (mm)23 5.23 5.25 5.20 5.20 4.09 5.26 5.26 5.26 5.26 2.88 5.28 3.81 5.22

Crack Growth (mm) 2.03 2.20 2.10 025 1.59 1.17 025 025 025 1.03 2.40 0.51 1.41

Pressure Range (barg) 18 to 32 10 to 29 10 to 29 18 to 32 10 to 29 17 to 29 10 to 29 17 to 29 10 to 29 17 to 29 10 to 29 21 to 29 10 to 29

Hoop Stress Range (MPa) 80 108.10 109.10 80.42 107.99 68.2 107.99 68.20 107.99 67.97 107.61 45.72 108.57

Bending Stress Range (MPa) 126.8 200.3 187.8 121.3 203.9 117.4 203.9 117.4 203.9 117.1 202.3 63.4 172.6

No. Of Cycles from BS 7910 1,524 623 616 0 1000 633 0 0 0 7,000 766 4,000 1,176

No. of Cycles from Test 12,024 5,305 3,800 12,460 1,000 7,000 3,500 7,000 1,810 7,000 2,735 4,000 275

Safety Factor24

7.9 8.5 >6

> 11

3.6 4.3

Table 4: Retrospective Studies on Fatigue Crack Growth.

22 23

From test. Predicted using BSI 7910. 24 No. of cycles from test / no. of cycles predicted by BSI 7910. 25 Predicted to fail.

10

Copyright 2012 by ASME

ANNEX B: FIGURES

long seam weld

Peak or roof top

Figure 1: An Illustration of the Seam Weld and Roof Topping Effect.

Figure 2: A Ring Tension Test Machine.

Figure 3: Failed Section of Pipe (2009).

11

Copyright 2012 by ASME

Figure 4: Class E S-N Assessment Curves.

12

Copyright 2012 by ASME

Class E, Mean Curve R24 (2.8mm) R19 (1.35mm)

Class E, -2 Std R10 (2.7mm) R15 (2mm)

Class E, +2 Std R13 (2.6mm) R18 (1.35mm)

R17 (4.25mm) R12 (2.55mm) R22 (2.7mm)

R11 (3.65mm) R23 (2.8mm)

1000

Stress,MPa (Log Scale)

100

10

1 1.E+02

1.E+04 N, Cycles (Log Scale)

1.E+06

1.E+08

Figure 5: Class E S-N Curves and Failures of Rings with EDM Slits.

13

Copyright 2012 by ASME

Class E, Mean Curve

Class E, -2 Std

Class E, +2 Std

R20 (No Defect)

Ring 25 (No Defect)

1000

100

Stress, MPa (Log Scale)

10

1 1.E+02

1.E+04 N, Cycles (Log Scale)

1.E+06

1.E+08

Figure 6: Class E S-N Curves and Failures of 'Defect Free' Rings.

14

Copyright 2012 by ASME

Class E, Mean Curve F8 - 1980 F6 - 1980

Class E, -2 Std F4 - 1980 J7 - 1980

Class E, +2 Std J1 - 1980

F1 - 1980 Test J4 - 1980

1000

100

Stress, MPa (Log Scale)

10

1 1.E+04

1.E+06 N, Cycles (Log Scale)

1.E+08

Figure 7: Class E S-N Curves and the 1980 Fatigue Tests.

15

Copyright 2012 by ASME

Class E, Mean Curve

Class E, -2 Std

Class E, +2 Std

1980 Failure

2009 Failure

1000

100

Stress, MPa (Log Scale)

10

1 1.E+03

1.E+05 N, Cycles (Log Scale)

1.E+07

Figure 8: Class E S-N Curves and the 1980 and the 2009 Failures.

16

Copyright 2012 by ASME

Class E, Mean Curve R24 (2.8mm) R12 (2.55mm) R18 (1.35mm) Ring 25 (No Defect) F4 - 1980 J7 - 1980

Class E, -2 Std R11 (3.65mm) R23 (2.8mm) R22 (2.7mm) R20 (No Defect) J1 - 1980

Class E, +2 Std R10 (2.7mm) R19 (1.35mm) 1980 Failure F1 - 1980 Test J4 - 1980

R17 (4.25mm) R13 (2.6mm) R15 (2mm) 2009 Failure F8 - 1980 F6 - 1980

1000

Stress, MPa (LogScale)

100

Failures + Rings with Initial Defects

'Defect Free' Rings

10

1 1.E+03

1.E+05 N, Cycles (Log Scale)

1.E+07

Figure 9: Fatigue Life Comparison.

17

Copyright 2012 by ASME

EDM Slits of 1.2 to 4.4 mm

Lamination (EDM: 2.5 mm)

Defect Free Ring (Roof Topping: 4.5 mm)

Defect Free Ring (No Roof Topping)


2009 Failure

Defect Free Specimens (1980's Fatigue Tests)

1980 Failure

Roof Topping Effect

No Roof Topping Effect

Rings with Pre-existing Defects + Failures

Defect Free' Rings

1.0

10.0

100.0

1000.0

10000.0

100000.0

No. of Years (Log Scale)

Figure 10: Operational Assessment of Fatigue Tests and Failures.

18

Copyright 2012 by ASME

Figure 11: View of the Fracture Surface of a Ring from Static Ring Tests.

Figure 12: Electron Micrographs of the Ring.

19

Copyright 2012 by ASME

700

Ring 21 (EDM Slit with Laminations)


Ring 7 (Fatigue Cycled before Burst Test, No Crack Initiated 600
Prediction from Industry Prediction from the Battelle Equation

EDM Slit only

500

Hoop Stress (MPa)

400

300

200

100

0 0 1 2 3 4
Depth of EDM Slit (mm)

Figure 13: Ring Test Results for Rings with No EDM Slit + with an EDM Slit.

20

Copyright 2012 by ASME

Ring 21 (EDM Slit with Laminations) Ring 7 (Fatigue Cycled before Burst Test, No Crack Initiated EDM Slit only

EDM Slits with Fatigue Cracks (Burst Tests) Prediction from the Battelle Equation Predicted from Industry

700

600
Hoop Stress (MPa)

500

400

300

200

100

0 0 1 2 3 4 5 6 7 8 9
Slit or Crack Depth (mm)

Figure 14: Test Results for Rings with EDM Slits and Fatigue Pre-cracks.

21

Copyright 2012 by ASME

Ring 21 (EDM Slit with Laminations) Ring 7 (Fatigue Cycled before Burst Test, No Crack Initiated Ring 20 (Fatigue Test, No Initial Defect)
Predicted from Industry Prediction fromthe Battelle Equation

EDM Slits with Fatigue Cracks (Burst Tests) Ring 25 (Fatigue Test, No Initial Defect) Ring 22 (Fatigue Tests with Lamination) EDM Slit only

Fatigue Tests 700

600
Hoop Stress (MPa)

500

400

300

200

100

0 0 1 2 3 4 5 6 7 8 9
Slit of Crack Depth (mm)

Figure 15: All Ring Test Results, with EDM Slits or Fatigue Cracks.

22

Copyright 2012 by ASME

You might also like