In-line inspection detects early cracking on Canadian crude-oil line

Susan E. Miller
Interprovincial Pipe Line Inc.
Edmonton

Michael A. Gardiner, Clive R. Ward
BG Technology
Loughborough Leicestershire, U.K.

• An elastic wave in-line inspection (ILI) tool has arrived at a pig-receipt station after a run to detect cracks along Interprovincial Pipe Line's Line 3 (Fig. 2). [52,683 bytes]
• A fatigue crack, like this one shown in cross-section from a 1989 Line 3 fatigue crack, can grow gradually over years and fail under normal line operation (Fig. 3). [45,470 bytes]
• Crews excavate a section of Line 3 to verify results of the EW-vehicle analysis (Fig. 5). [58,438 bytes]
• Defects repaired by sleeving in 1996 [21,166 bytes]

A program of in-line inspection (ILI) in 1996 by Interprovincial Pipe Line Inc. (IPL), Edmonton, established the integrity of one particular line segment before it was hydrotested.

Several defects were identified and repaired, but only one may have been large enough to have failed the hydrotest. At the same time, the lack of any failures during the hydrotest demonstrated that ILI is reliable and overlooks no defects that would have been critical up to 100% specified minimum yield strength (SMYS).

The work afforded the opportunity to compare results from ILI for cracking with a hydrostatic retesting, scheduled weeks after locations pinpointed by the inspection were excavated.

IPL is a wholly owned subsidiary of IPL Energy Inc. Together with affiliate Lakehead Pipe Line Partners L.P., Duluth, Minn., it operates the world's longest liquid hydrocarbon pipeline system.

This system extends 5,100 km from Edmonton to Superior, Wis., and Montreal (Fig. 1 [87,540 bytes]). It delivers an average of 1.7 million b/d of liquid petroleum from western Canadian producers to refining centers and markets in eastern Canada and Midwestern U.S.

Inspection

The BG elastic wave (EW) in-line crack detection vehicle was used to inspect 213.5 km (133 miles) of IPL's Line 3 which runs from Edmonton to the international boundary near Gretna, Man. At this point, it makes an end-on connection with the Lakehead system for further transmission to Superior.

Rigorous analysis of the inspection data, concentrating on the seam weld and surrounding region, identified 73 sites for excavation.

Pressure-retaining sleeves were fitted at 17 locations. Of these, the most severe defect was a 25 mm (1 in.), 40% through-wall, long-seam shrinkage crack. This was the only feature that might have failed under hydrotest to 100% SMYS.

Twelve other cracks, each measuring 20-35% through wall, were sleeved. Minor imperfections were found at the majority of others reported but were not sleeved.

Following completion of remedial work, 198 km of Line 3 were hydrostatically tested at pressures up to 100% SMYS, including 156 km that had been inspected by EW vehicle.

There were no leaks or ruptures under hydrotest, demonstrating the ability of the tool reliably to detect cracks in the seam weld and surrounding region that were smaller than would have been found by hydrotesting alone.

BG Technology developed the EW crack-detection vehicle (Fig. 2), which is owned and operated by Pipeline Integrity International Ltd., Cramlington, U.K. BG Technology, a unit of BG plc, Reading, U.K., conducts research and development in all parts of the gas supply chain, from exploration to burner design.

The EW vehicle detects cracks by transmitting circumferential elastic shear waves at ultrasonic frequencies into the pipe wall. These are generated by transducers in liquid-filled wheels, coupled to the inside pipe surface through soft tires. Spring loading maintains good contact with the pipe wall, which allows the system to be used without slugs in gas pipelines as well as in liquids.

Mann gives further details of the instruments and processes used by the vehicle.¹ Johnston and Thomas² and Ward³ describe operational experience with the tool.

It has also been shown that the EW tool can detect and characterize defects that were subsequently shown in laboratory burst tests to sustain pressures equivalent to as much as 133% SMYS before failing.⁴

The inspection vehicle used to perform the work described here is known as the Mark 2. It deploys 32 transducer wheels, operating in 16 clockwise-counterclockwise pairs, and fits pipe from 762 to 914-mm (30 to 36-in.) OD. Onboard data recording uses a reel-to-reel tape recorder, and the range is typically 45-50 km, depending on the ultrasonic characteristics of the pipe steel.

Line 3

Line 3 of the IPL system is an 864 mm (34 in.) OD pipeline built between 1962 and 1968 with predominantly X-52 Grade DSAW line pipe, 7.1-mm (0.281-in.) W.T. Protection against the ground environment is by single-layer polyethylene tape wrap.

The line transports crude oils of varying density from sweet, light crude to heavy crude with a viscosity of 350 cSt.

The section of Line 3 inspected lies between the tool launcher at Regina, Sask., and receiver at Cromer, Man. This section is 253 km and has three intermediate pump stations at Odessa, Glenova, and Longbank with an average separation of 57 km between stations.

In 1989, Line 3 experienced an in-service rupture downstream of the Langbank, Sask., pump station. This was found to have been caused by fatigue cracking at the base of some light external corrosion (less than 8% through wall), aligned with and close to the DSAW long seam toe.

Examination of the excavated pipe joint found no evidence of metallurgical or structural weakness, and it was concluded that the corrosion had created a concentrator for in-service stress variations. Over many years, this had allowed a fatigue crack to grow so that the joint would fail in normal operation (Fig. 3).

The 1989 failure caused IPL to evaluate the potential for finding existing subcritical cracks by hydrostatic retesting. IPL concluded that, even leaving aside the logistical problems and throughput impact of hydrotesting, nondestructive in-line inspection was potentially far better.

The company based this conclusion on the ability of ILI to find much smaller defects than was possible with hydrotesting. By finding defects at sizes down to early growth phase, IPL would be able to assess the defect population and build up the most effective long-term, risk-management strategy.

Accordingly, IPL began to develop an in-line crack-detection program, including research to study the effects of operating the EW tool in a liquid line and a series of runs with the 864-mm (34-in.) EW tool.

Between June 1995 and February 1996, Line 3 experienced three in-service failures in the Regina-to-Cromer segment. The first was attributed to a unique shape of external corrosion that rendered its size difficult to predict with normal, high-resolution magnetic-flux-leakage (MFL).

The second occurred within 1 km of the 1989 failure and again was found to be a fatigue crack in minor (8% through wall) corrosion at the long-seam toe.

The third failure resembled the first case of elongated corrosion with the addition of a 15% through-wall externally initiated stress corrosion crack at the base of the corrosion.

Following this failure, IPL reached agreement with Canada's National Energy Board (NEB) voluntarily to reduce pressure by 20% on Line 3 between Odessa and Cromer and to conduct an enhanced program of integrity re-evaluation.

The program would ultimately be evaluated and the normal operating pressures restored, pending successful hydrostatic retesting of the 198 km of line where the failures had occurred.

As well as the known problem of fatigue cracking and the potential for general corrosion and environmentally assisted cracking (EAC) under disbonded tape wrap, IPL during 1995 found examples of a novel form of corrosion which came to be known as "narrow axial external corrosion" (NAEC).

It consisted of corrosion grooves with lengths several orders of magnitude greater than their widths. These features were typically very narrow compared to general corrosion, perhaps a few millimeters in width at most.

NAEC often turned up at the toe of the external long seam, initiating and growing under the narrow tent of disbonded polyethylene tape that was an artifact of installation over the long-seam profile. Groundwater could then accumulate in the "tenting" caused by the weld bead under tape wrap.

These corrosive conditions would, of course, also encourage development of EAC, while the corrosion groove itself would be a stress concentrator that could eventually lead to fatigue failure such as at Langbank.

A particular problem with NAEC is that its morphology makes it hard to identify unambiguously and to size by longitudinal field MFL methods as are widely used for metal-loss ILI surveys. Similarly, it is impossible for circumferential EW technology to quantify the severity of such features, although they can be detected.

The discovery of these three separate yet related phenomena (NAEC, cracks inside NAEC, and long-seam fatigue) on Line 3 meant that IPL faced some challenges preparing for the hydrotest. There was also a throughput imperative to achieve a successful hydrotest as soon as possible.

But before hydrotesting could be done, it was necessary to locate and repair all defects that could be critical at the test pressure.

Regina to Cromer

All of Line 3 was inspected for metal loss by Pipeline Integrity International's high-resolution MFL vehicle in 1989-1990 and reinspected in 1993-1994. The Mark 2 EW crack-detection vehicle had made several runs between Regina and Cromer in 1994 and 1995.

Based on analysis of the 153 km of EW data collected then, IPL excavated 22 pipe joints with EW indications. Of these, 7 had crack indications either in the external toe of the long seam, externally in the pipe body, or internally in the center line of the long seam. Table 1 [23,190 bytes] lists the range of defect classifications from the 1994-95 program.

Fig. 4 [66,010 bytes] shows a schematic of the 1994-95 inspections, together with the crack-detection runs made in 1996. In summary, the 1994-95 program had collected data from 153 km, including some short overlaps between runs.

IPL knew, however, that data from some of these runs had been degraded by mechanical damage to the inspection vehicle's transducer wheels. It had been understood that, given the transducers' paired operation, the derangement of one wheel of a pair would not prevent detection and discrimination of a potentially injurious feature.

After early successes in applying the EW technology to Line 3, however, IPL traced the 1995 Langbank failure to a defect which, although present in inspection data, had been incorrectly classified. An investigation indicated that this was attributable to one wheel of the detecting pair having malfunctioned.

Following this discovery, engineers at BG Technology immediately re-examined the criteria for acceptable data quality, particularly in the seam weld and nearby regions. They generated new data-quality criteria and devised an improved system of checks. In addition, they modified the processes for interpreting data from the seam weld and areas close by. These improved processes are now part of the routine data-analysis procedure.

Working closely with IPL, specialists at BG Technology began 1996 by reanalyzing all existing EW data from Line 3, using the new methods.

This allowed a quantitative picture of the existing data quality to be built up, with particular reference to the seam-weld region considered particularly susceptible to fatigue cracking. With this information, the specialists formulated a program of crack-detection runs for 1996.

New runs

A 1996 program of ILI runs, conducted ahead of hydrostatic testing, used several technologies. The program used both MFL and ultrasonic tools to size metal loss, as well as the EW tool to detect longitudinal cracks.

The EW target inspection ranges were designed to optimize coverage of the seam-weld region between Regina and Cromer. Five runs in the Regina-to-Cromer area took place between Apr. 15 and May 4, 1996.

It should be noted that use of multiple runs was a function of the Mark 2 vehicle's range. The Interim Mark 3 vehicle now in service has a range of up to 150 km/launch.

Data from 176 km, when combined with previous years' inspections, give coverage of 213.5 km out of the 253 km between Regina and Cromer. Fig. 4 shows the coverage from the runs.

Reanalysis of existing data had shown that the seam-weld region was completely covered for the 17.1 km immediately downstream of Regina and for 46.4 km immediately downstream of Odessa. Accordingly, no further inspection of these sections was warranted.

The section from 17 km downstream of Regina to Odessa was programmed for inspection, having never been surveyed previously. In the program, a single pass collected good data between 16 km downstream of Regina and 9.1 km downstream of Odessa.

IPL had concluded that the existing data from Glenavon to Langbank was badly degraded by mechanical damage to the pig, and reinspected this section. Two runs resulted in good-quality data for 49.7 km, starting 6.8 km upstream from Glenavon.

Finally, two runs timed to begin recording slightly before Langbank yielded good seam weld data for 46.9 km, commencing 4.1 km upstream of the Langbank pump station.

While BG Technology and IPL were establishing the program of runs for the crack-detection vehicle, Pipeline Integrity International was addressing the problem of detection of NAEC and cracks inside NAEC.

The details of that approach lie outside the scope of this discussion, but a new MFL tool was developed and used in Line 3 early in April 1996. IPL incorporated results from this new tool in the analysis of EW data, described presently.

Data analysis, cross referencing

BG Technology analyzed new data and reanalyzed previous records and, in this process, used several novel approaches to account for the line's unique characteristics and to make best use of all available data from the various technologies.

Regarding the line's characteristics, IPL agreed that the seam-weld region (defined as 100 mm either side of the long seam, together with the seam itself) was critical. Technicians concentrated analysis on this region, except for joints showing evidence of NAEC or other axially aligned metal loss. They examined the full pipe body for cracking in such joints.

Wherever overlaps existed between runs of the crack-detection pig, the best-quality data for the region of interest were used for primary analysis. Other EW data corroborated this analysis.

Technicians extensively cross-referenced data from all available technologies to build up a complete and accurate picture of the line's condition. For example, when a crack-like feature was flagged in the EW data, IPL checked the MFL metal-loss data at the feature's location.

This allowed compensation for the effects of corrosion to be factored in to the feature's assessment and also ensured that the combined depths of metal loss and cracking would be reported.

Likewise, if the new MFL tool detected any axial corrosion, IPL checked for coincident crack-like indications in the EW record, in case stress concentration had led to fatigue or stress corrosion cracking within the metal loss.

Field excavations; retesting

IPL learned of excavation sites according to an agreed schedule that allowed field crews to be effectively used from April to September 1996. Technicians identified sites by reference to pipeline features and girth welds, from which GPS co-ordinates were derived for accurate location in the field.

Once a site had been uncovered, independent third-party nondestructive evaluation (NDE) specialists examined the reported area of pipe wall and reported.

IPL received reports on 73 sites, all of which were excavated (Fig. 5). Ultrasonic indications were found at all but 5 of these sites and pressure-retaining sleeves were applied at 17 locations (Table 1).

The largest crack found, and the only one which may have failed hydrotesting to 100% SMYS, was a 40% through-wall, 25-mm (1-in.) internal long seam shrinkage crack. Three other sleeved sites had shrinkage cracks between 30% and 35% through wall, while all other sleeved features were less than 25% through wall.

Most of the 56 unsleeved sites were found to have noninjurious features such as cold laps at the long seam or inclusions within the plate. Other indications were from minor features with depth less than 10% through wall and were either ground out, if external, or left without further action.

There had been concern before the inspections that there may have been instances with cracking combined with metal loss, which had motivated the data cross-referencing previously described. In fact, excavation turned up no such defects.

While data analysis and in-field repairs were going on, IPL was also preparing to hydrotest Line 3 between Odessa and Cromer, as required by the NEB.

Major construction was necessary to allow for the intake of 140,000 cu m of water upstream from Regina and also the building of a $5.2 million (Canadian) water-retention pond at Cromer to allow for water cleansing and disposal. This work was completed in time for the hydrotest to begin on Sept. 16, 1996.

IPL designed the hydrotest to include a strength test of 4 hr at a maximum of 860 psi, to be followed by a 4-hr leak test at 120% MAOP. This program was applied concurrently to the eight sections into which the line had been divided by block valves between Odessa and Cromer.

The entire hydrotest was successfully completed within 30 hr of commencing pressurization, with no ruptures or leaks being experienced. The line soon returned to service and, shortly afterwards, IPL received approval to remove the 20% operating-pressure restriction from the section.

Developments

There have been major advances in EW technology since the inspection program reported here.

The Interim Mark 3, as a precursor of the Mark 3 vehicle, which will be introduced later this year, replaced the Mark 2 device early in 1997.

The Interim Mark 3 uses 64 transducers, which may be all of one type to give redundant coverage or of two types to allow the use of different sensor types for improved feature discrimination. Despite the extra transducers, the range was increased to a maximum of approximately 150 km by using new hardware and software. This tool is available in the same range of sizes as the Mark 2.

The Mark 3 vehicle is being developed under a US$5.4 million program sponsored by Canadian Energy Pipeline Association, Gas Research Institute, Pipeline Research Committee International, and Pipeline Integrity International. This development will continue the improvements made in the Interim vehicle by having longer range, up to 96 transducers, variable product bypass, and sizes from 508 to 1,219 mm (20 to 48 in.).

Acknowledgments

The authors thank the group director for research and technology at BG plc and the management of Interprovincial Pipe Line Inc. for permission to publish this article.

References

1. Mann, A.S., Ward, C.R., Paige, D., and Lazor, R., "Inspecting Liquids Pipelines For Longitudinal Cracking," Second International Conference on Pipeline Technology, Ostend, 1995.
2. Johnston, D., and Thomas, S., "Colonial's Experience with Finding Longitudinal Defects with Internal Inspection Devices," First International Pipeline Conference, Calgary, 1996.
3. Ward, C.R., Dunford, D.H., and Mann, A.S., "Inspecting Operational Pipelines for Stress Corrosion and Fatigue Cracking," Institute of Gas Engineers, London, 1993.
4. Maxey, W.A., Mesloh, R.E., and Kiefner, J.F., "Use Of The Elastic Wave Tool to Locate Cracks Along The DSAW Seam Welds In A 32-Inch OD Products Pipeline," Second International Pipeline Conference, Calgary, 1998.

The Authors