ANALYSIS OF SEAM WELD RELATED PIPELINE FAILURES

G. T. Quickel, B. C. Rollins, and J. A. BeaversCC Technologies, Inc.

5777 Frantz RoadDublin Ohio 43017-1386 USA

Phone (614) 761-1214 Fax (614) 761-1633Gregory.Thomas.Quickel@dnv.com

Keywords: SSAW, ERW, fatigue, hook, groovingAbstract

There are more than 2.5 million miles of oil and gas pipelines in the United States. These pipelines typically contain longitudinal seam welds in each pipe joint and girth welds that connect the individual joints to form the pipeline. Both types of welds are prone to failure from time independent and/or time dependent failure mechanisms. This paper reviews common pipeline welding types and provides several case histories in which failure analysis techniques were used to determine a metallurgical cause of failure.

Introduction

While some smaller diameter pipelines are seamless, most pipelines are manufactured by forming flat plate or skelp into a tubular form and completing a longitudinal seam weld. Both submerged arc welding and autogenous welding processes are used for weld completion.

Submerged arc welded line pipe is manufactured by first forming a flat plate or skelp into a tubular shape (can) in a set of presses, followed by weld completion.(1) Prior to forming the can, the edges are typically beveled. Historically, single submerged arc welding (SSAW) and double submerged arc welded (DSAW)(2) processes have been used but, currently, the DSAW process is the only submerged arc welding process that is approved in API 5L.(3) In SSAW line pipe, the edges are joined by a single pass submerged arc weld made from the outside surface onto a backing shoe located at the ID surface. DSAW line pipe is formed in a similar manner except one pass is made from the OD surface followed by a pass from the ID surface, or vise versa. Filler weld material is used in both processes. One variation of this process is used to producespiral welded DSAW line pipe; in which skelp is helically wound and welded to produce a spiral weld.

Historically, there have been several different autogenous welding processes for longitudinal seam welds including furnace lap welding, furnace butt welding, electric flash welding (EFW) and electric resistance welding (ERW).(4) ERW currently is the dominant autogenous welding process for pipe manufacturing. ERW line pipe is manufactured by forming plate or skelp into a tubular shape and heating the two adjoining edges with electric current and forcing them together mechanically. An autogenous bond is formed between the molten edges. Upset material at the weld is trimmed on the OD and ID surfaces.

Materials Science and Technology (MS&T) 2008October 5-9, 2008, Pittsburgh, Pennsylvania • Copyright © 2008 MS&T’08®

513

Failure Analysis for Problem Solving

Various types of defects can be produced in these welds and the defects typically are unique to the specific welding procedure. Some of these defects are too small to be detected in the mill and are never an integrity problem for the pipelines. Other defects that are not detected at the mill can fail during the initial hydrostatic test of a pipeline, or grow in service by fatigue, stress corrosion cracking, or other mechanisms, resulting in a service leak or failure. Because of differences in the metallurgy at the weld and the base metal of the pipe, the welds also can be prone to environmentally induced failure mechanisms such as preferential corrosion. This paper describes three analyses of pipeline failures that initiated at the longitudinal seam welds from these processes.

Case History No. 1 – Grooving Corrosion Failure

Background

A metallurgical analysis was performed on a section of line pipe that leaked at the seam weld during a hydrostatic pressure test. The portion of the pipeline that leaked was comprised of 12.75-inch diameter by 0.375-inch wall thickness, Grade B line pipe that contained an ERW longitudinal seam. The pipeline transported refined petroleum products.

The pipeline was installed in 1953 and was externally coated with coal tar. The coating was disbonded on the entire pipe section and liquid and corrosion product were present underneath the disbonded coating. The pipeline had a sacrificial anode cathodic protection (CP) system.

The maximum operating pressure (MOP) for this pipeline was 275 psig, which corresponds to 13.4% of the specified minimum yield strength (SMYS). For the hydrostatic pressure test, the pipeline was initially pressurized to 258 psig (12.8% of SMYS) and held for one hour. The pressure was then raised to 410 psig (19.9% of SMYS) where it dropped 15 psig for every five minutes, which indicated the presence of a leak.

Approach

The pipe section was visually inspected and photographed in the as-received condition. Scale was removed from the external surface and elemental analysis using energy dispersive spectroscopy (EDS) with a scanning electron microscope (SEM) was performed. Portions of a groove at the seam weld were removed, placed in liquid nitrogen, hit with a hammer to expose the fracture surfaces, and photographed. Transverse cross-sections were removed from the exposed fracture surfaces and in-tact seam weld, mounted, polished, and etched. Light photomicrographs were taken to document the seam weld morphology and microstructure. Samples were removed from the fracture surfaces, cleaned in inhibited acid with a soft bristle brush, and examined in the SEM to document the morphology of the fractured and corroded surfaces. Steel samples were removed from the pipe section for chemical composition analysis.

514

Results

The results of the analysis indicated that the leak occurred as a result of grooving corrosionat the ERW longitudinal seam. External corrosion was present in the vicinity of the groove that was located at the seam, Figure 1. The groove was approximately 8.75 inches in length and through-wall for approximately 1/8 inch, Figure 2. As shown in Figure 2, Region 1 is located adjacent to the OD surface and contains brown oxide. This is the pre-existing groove at the seam. Region 2 is located adjacent to the ID surface, has a grey and shiny appearance, and was formed after breaking open the pipe sample. EDS analysis of the deposits (chlorine detected) in the groove was consistent with a corrosive environment.

Figure 1 Photograph of the external pipe surface along the seam weld in Case History № 1.

Figure 2 Photograph of the fracture surface showing the through-wall leak location in Case History № 1.

Metallurgical analysis of a cross-section (Figure 3) through the groove revealed that 1) the groove was V-shaped and 2) that the apex of the groove was located at the bond line (white vertical stripe) of the ERW seam. Both of these observations are consistent with a grooving mechanism that is known to occur in ERW and EFW seams. ERW and EFW seams in the presence of a corrosive environment are susceptible to preferential attack at the bond line. The material at the bond line is anodic to the surrounding material and the result is a V-shaped groove with the apex of the V centered on the bond line.

OD

ID

Through-wall Region 1 Region 2

515

Figure 3 Light photomicrograph of a cross-section removed from the seam weld away from the through-wall leak location in Case History№ 1.

V-shaped groove, Region 1

Bond Line

OD

ID

516

The microstructure of the heat affected zone (HAZ) of the weld consisted of ferrite grains in a matrix of spheroidized pearlite. The microstructure of the base metal consisted of inclusions and equiaxed grains of ferrite and pearlite. The microstructure of the bond line consisted only of ferrite grains. The variation in microstructure between the base metal and HAZ may create minor localized galvanic differences, with the seam structure being more anodic. The microstructure and chemistry of the steel was typical of the vintage and grade and the chemistry met the API 5L specifications in place at the time of manufacture.

Case History No. 2 – Hook Crack Failure

Background

A metallurgical analysis was performed on a section of line pipe that ruptured at the seam weld during a hydrostatic pressure test. The portion of the pipeline that failed was comprised of 16-inch diameter by 0.312-inch wall thickness, API 5L X52 line pipe that contained an ERW longitudinal seam. The pipeline transported refined petroleum products. The MOP on this line segment was 1408-psig, which corresponds to 69.4% of SMYS. The failure occurred during initial pressurization at a test pressure of 1390-psig, which corresponds to 98.7% of the MOP and 68.5% of the SMYS. The normal operating pressure at the failure location ranged from 1000 to 1100-psig (71.0 to 78.1% of MOP).

The pipeline was installed in 1965 and was externally coated with coal tar. The coating was reportedly intact near the failure. The pipeline had an impressed current CP system that was commissioned around 1965. This pipeline segment was previously hydrostatically tested in the fall of 1965. The hydrostatic test lasted twenty-four hours and the maximum pressure was 1760-psig (125% of MOP and 86.8% of SMYS).

Approach

The pipe section was visually examined and photographed in the as-received condition. Transverse base metal and cross-weld samples were removed from the pipe section for mechanical (Charpy V-notch and duplicate tensile) testing. Samples for chemical analysis of the steel were removed from the base metal. Magnetic particle inspection (MPI) was performed where the coating was removed to identify defects at or near the seam weld. Transverse metallographic samples were removed from the seam, at and away from the failure origin. The samples were mounted, polished, and light photomicrographs were taken to examine the morphology and steel microstructure. Samples were removed from the failure origin to analyze the morphology of the fracture surface in the SEM.

Results

The results of the analysis indicated that the rupture initiated at an ID connected pre-existing hook crack. This and all hook cracks are slightly offset from the bond line of the ERW seam, Figure 4. No evidence of in-service growth by fatigue was found, although the quality of the fractography was poor as a result of corrosion of the fracture surfaces that occurred after the ruptures. The tensile properties of the line pipe steel and the steel chemistry were typical for the vintage and grade and met the API 5L specifications in place at the time of manufacture. The

517

microstructure and Charpy toughness properties of the steel also were typical for the vintage and grade.

Figure 4 Light photomicrograph of the cross-section removed from the failure origin in Case History № 2.

518

There was no evidence of external corrosion of the pipe section, which indicates that the coating was intact prior to the rupture and was removed by the rupture or by subsequent handling. The pre-existing hook crack that was surface breaking on the ID surface was evident on the fracture surface, Figure 5. The hook crack was approximately forty-four inches in length, with a maximum depth of 40% of the wall thickness.

Figure 5 Photograph of a portion of the fracture surface in Case History № 2.

Hook cracks can form during welding when metal separations that are parallel to the pipe surface are present.(5) The parallel separations form “hooks” when the edges of a pipe are upset during welding. As shown in this case history, failures from hook cracks can and have occurred during subsequent pressure testing of older ERW pipelines. Failures can also occur when fatigue cracks propagate from hook cracks. Hook cracks are known to also occur in EFW seams.

Case History No. 3 – Lack Of Fusion DefectBackground

A metallurgical analysis was performed on a section of line pipe that ruptured at the seam weld while transporting crude oil. The portion of the pipeline that failed was comprised of 20-inch diameter by 0.344-inch wall thickness line pipe with a SMYS of 52 ksi and contained a SSAW longitudinal seam. The MOP on this line segment was 1098-psig and the average operating pressure was 757 psig. The pressure at the time and location of failure was 1,052 psig.

The pipeline was constructed in 1952, was externally coated with coal tar, and contained an impressed current CP system. The coating around the failure origin was intact and did not visually appear to be disbonded.

OD

ID

FinalFracture

Pre-existinghookcrack

519

A hydrostatic test was performed in 1992 on this line segment. The test pressure at the location of the failure was calculated to be 1,421 psig. Magnetic flux leakage (MFL), transverse field inspection (TFI), and ultrasonic crack detection (USCD) in-line inspection (ILI) surveys were previously performed on this line segment. A Petrosleeve was installed in February 2002 just upstream (U/S) of the failure origin in response to the investigation of two (2) TFI features that were measured to be 32% and 34% of the wall thickness.

Approach

The pipe section that contained the failure origin was visually examined and photographed in the as-received condition. The coal tar coating was removed with a brass hammer, methyl ethyl ketone (MEK), and a brass bristle brush at the seam weld. MPI was performed on the external surface of the in-tact seam weld two feet U/S and downstream (D/S) from the ends of the rupture opening. Following MPI, the external surface was photographed.

The fracture surfaces were removed, optically examined, and photographed. The length and depth of the flaw present on the fracture surface were measured to produce a flaw profile. One cross-section was removed U/S of the rupture opening at an indication, three cross-sections were removed adjacent to the rupture opening, and another cross-section was removed D/S of the rupture opening at indications. The cross-sections were mounted, polished, and etched. Light photomicrographs were taken to document the steel microstructure and morphology of the fracture or indication. Hardness testing was performed across the weld on a mounted cross-section. Three samples containing the fracture surface were removed from the rupture opening and analyzed with a stereo microscope and SEM. Samples for chemical analysis were removed from the weld metal and base metal to determine the composition. Transverse tensile and Charpy V-notch impact specimens were removed from the weld metal and HAZ, respectively, of the pipe section.

Results

The results of the analysis indicated that the rupture initiated at an ID surface breaking lack-of-fusion (LOF) defect that was located at the root of the SSAW seam, Figure 6. The LOF defect was one of three regions observed on the fracture surface at the failure origin. The three regions shown in Figure 7 were: a smooth and relatively featureless region near the ID surface that was the LOF defect (Region 1), a mid-wall region that contained fatigue striations(Region 2, Figure 8), and an overload region produced by the final failure (Region 3). Region 3 was ductile in nature at the deepest portion of the flaw and located adjacent to the OD surface. The primary flaw, which contained Region 1 and 2, was approximately six inches in length with a maximum depth of approximately 0.302 inches (88.8% of average wall thickness), measured from the ID surface. An ID connected flaw, which a majority was just Region 1, was identified on either side of this primary flaw and extended approximately 3.4 feet in length with an average depth of about 30% of the average wall thickness.

520

Figure 6 Stereo light photomicrograph of the cross-section removed from the center of the pre-existing flaw in Case History № 3.

Figure 7 Photograph of the clockwise side of the fracture surface at the failure origin in Case History № 3.

OD

ID

Region 1

Region 2

Region 3

OD

ID

Region 2

521

Figure 8 SEM image of the fracture surface in Region 2 for Case History № 3.

Additional indications were found by means of MPI on both sides of the external surface of the seam weld. The lengths of the indications varied from 0.13 feet (1.6 inches) to 2.20 feet. There was no evidence of external or internal corrosion on the pipe section and the coating visually appeared to be intact. The pipe steel chemistry and tensile properties of the samples removed from the pipe section met the minimum composition and tensile specifications for API 5LX line pipe steel at the time of manufacture, respectively. The microstructure of the samples removed from the pipe section consisted of ferrite and pearlite, with some banding.

LOF defects occur when an inadequate bond is formed during the welding process. Reasons vary from insufficient heat and/or pressure, chemistry of the weld (carbon and sulfur content), oxidation, a narrow temperature range to fuse in forged seams, etc. LOF defects can occur in any type of weld in line pipe steels.

Fatigue is a common problem in any industry where cyclic loading is present. Liquid lines (refined products) have a greater propensity than natural gas pipelines when it comes to fatigue failures. A majority of fatigue cracks in seam welds develop at LOF defects but they also initiate from hook cracks and other defects or originate at the OD or ID surfaces. Fatigue can occur in any type of seam weld. Three of the features commonly seen on fracture surfaces associated with fatigue are ratchet marks, beach marks, and fatigue striations. Ratchet marks are parallel to

FatigueStriations

522

the direction of crack propagation and can be used to determine the location of fatigue initiation.(6) Beach marks are macroscopic marks on a fracture surface that indicate successive positions of the advancing crack front. They typically appear as semi-elliptical lines radiating from the origin of the crack; thus, they can be used to determine the crack initiation location. Fatigue striations are microscopic marks on a fracture surface that also indicate successive positions of the advancing crack front. Unlike beach marks, they typically are associated with individual load cycles, and therefore are much more closely spaced than beach marks.

CONCLUSIONS

The pipeline infrastructure continues to age and many pipelines are operating well past their original design lives. Safe operation of these pipelines can be maintained with a well planned and executed integrity program and such a program should address the potential for pre-existing weld related defects to grow in service or for defects to initiate and grow at the welds. One component of a well executed integrity management plan is proper sentencing of the defects that cause failure, by means of a metallurgical failure investigation. This paper provided examples of several different types of seam weld defects that can compromise pipeline integrity.

REFERENCES

1. J. F. Kiefner and E. B. Clark, “An ASME Research Report: History of Line Pipe Manufacturing in North America,” CRTD-Volume 43, 1996.

2. P. G. Fazzini, J. C. Belmonte, M. D. Chapettie, and J. L. Otegui, “Fatigue Assessment of a Double Submerged Arc Welded Gas Pipeline,” International Journal of Fatigue, 291115-24 (2007).

3. API Specification 5L, 42nd edition, July 1, 2000.

4. Lu Shuanlu, Han Yong, Qin Changyi, Yuan Pengbin, Zhao Xinwei, and Luo Jinheng, “Crack and Fitness-for-Service Assessment of ERW Crude Oil Pipeline,” Engineering Failure Analysis, 13 565-71 (2006).

5. M. D. Chapettie, J. L. Otegui, and J. Motylicki, “Fatigue Assessment of an Electrical Resistance Weld Oil Pipeline,” International Journal of Fatigue, 24 21-28 (2002).

6. ASM: “Principles of Failure Analysis: Glossary of Failure Analysis Terms.”

523