Burke Delanty, John O'Beirne
TransCanada PipeLines Ltd.
Calgary
A 6-year field study by TransCanada PipeLines Ltd. (TCPL) of stress corrosion cracking (SCC) on its system has yielded important information about the origins and occurrence of SCC.
The study was prompted by three 1986 in-service line failures attributable to SCC.
Whether during service or TCPL's subsequent testing, SCC failures have occurred only in Class 1 locations (little or no human population). More populated areas on TCPL's system are protected by use of heavier wall pipe which consequently has greater flaw tolerance.
SCC, which in some cases led to both in-service or test ruptures, was detected under both polyethylene tape coated and asphalt-coated pipeline sections. Nevertheless, the study revealed it is much more prevalent under tape coating than asphalt coating. The study also found SCC in a line that has been in service for only 6 years.
SCC occurs in low pH (
But the vast majority of the SCC detected to date has reduced less than 10% of the wall thickness on TCPL's system, the study revealed; thus, TCPL's system is felt to be in no immediate danger. And SCC appears to be prevented when the pipe surface is adequately protected by a cathodic-protection (CP) system.
TCPL has learned that hydrostatic testing successfully prevents in-service SCC failures and will continue to be part of TCPL's strategy even as additional research progresses.
Moreover, TCPL's current practices for coating, cathodic protection, and design of new pipelines effectively prevent SCC.
LINE FAILURES
TCPL's natural-gas pipeline system consists of more than il,000 km (6,831 miles) of large-diameter pipe and 51 compressor stations from Alberta to Quebec. The pipeline is sectionalized by isolation valves at approximately 30-km (18.6-mile) intervals.
Between 1985 and 1986, TCPL's 914-mm (36-in.) line in northern Ontario experienced three in-service failures attributable to external stress-corrosion cracking (SCC).
The failures occurred at locations quite remote from each other and in Class 1 locations (low or zero population density), but each was located within 23 km (14 miles) downstream of a compressor station.
The line which failed was constructed in 1972 and 1973 from 9.1-mm (0.360-in.) W.T. Grade 448 (X-65) pipe and has a maximum allowable operating pressure (MAOP) of 6,895 kPa (1,000 psi) for most of its length and 6,455 kPa (935 psi) in the remaining portion. Most of the line operates in regions with little or no human population; in locations where population density warrants, heavier wall thicknesses have been used.
The earlier constructed sections of the line were coated with asphalt coating, while the later sections had field-applied polyethylene tape coating.
One of the failures occurred in an asphalt-coated section, while the other two were in tape-coated sections.
The SCC failure in the asphalt-coated section was associated with light, 0.18-mm deep scratches on the pipe surface. The two SCC failures in tape-coated sections initiated at the toe of the longitudinal seam weld (Fig. 1) and were not associated with any mechanical damage or metallurgical anomaly.
In response to these failures and to ensure safe and reliable service, TCPL initiated an extensive field program in the spring of 1986 aimed at controlling and understanding the long-term consequences of stress corrosion.
FIELD PROGRAM
The pipeline industry's experience with SCC had established a correlation between failure rates and operating temperature indicating that the greatest risk of failure lay in those sections located immediately downstream of compressor stations.
Since this pattern was consistent with TCPL's three in-service failures, it was initially decided to direct the majority of the field investigation and testing program towards the 28 valve sections of the 914-mm line that were situated immediately downstream of the stations between Winnipeg and Toronto.
It became apparent as the program progressed, however, that the type of SCC experienced by TCPL did not necessarily follow this temperature correlation and the program was broadened to include other sections of the line. Limited investigation was also conducted on other lines in the system.
The field program included hydrostatic testing of selected pipeline sections and conducting investigative excavations. It had three main objectives:
- To ensure the integrity of the 914-mm line
- To determine the extent to which this line and others in TCPL's system are affected by SCC
- To define the environmental conditions under which SCC is known to exist.
To date, the 28 pipeline sections immediately downstream of compressor stations between Winnipeg and Toronto and 20 additional sections (selected based upon population density, proximity to sections which failed in service, and environmental considerations) were each hydrostatically tested.
This test would rupture flaws significantly smaller than those which would fail at operating pressures and thereby ensures their integrity.
HYDROSTATIC TESTING
Each hydrostatic test consisted of pressuring the section to a hoop stress equivalent to 110% of the specified minimum yield strength (SMYS) of the pipe for I hr followed by a 23-hr leak test at a pressure of 110% of the MAOP or 7,585 KPa (1,100 psi).
The test pressure of 110% of SMYS or 9,655 KPa (1,400 psi) was chosen so as to remove the smallest flaws possible without causing general yielding of the pipe, while the 1-hr test duration reduced the possible growth of subcritical flaws during the test.
The leak test involved monitoring the pressure and temperature of the test section so as to identify any leaks which may have begun during the high-pressure test. The stress level associated with the leak test was unlikely to cause further growth of any remaining stress-corrosion flaws.
The initial testing program which commenced in 1986 resulted in four polyethylene tape-coated sections and two asphalt-coated sections experiencing a combined total of 16 failures during test.
Of the failures, nine were caused by stress corrosion, six by pitting corrosion, and one by a gouge associated with a dent. The nine SCC failures occurred in the four polyethylene tape-coated sections; the six corrosion failures occurred in both tape-coated and asphalt-coated sections.
Three of the valve sections that failed due to SCC were located within 30 km downstream of a compressor station, while the fourth valve section was surprisingly located 30-60 km downstream of a compressor station. This result contrasted sharply to the initial assumption that the valve sections downstream of compressor stations were likely to be the most susceptible to SCC failures.
One of the valve sections that failed on test due to SCC had also previously suffered a service failure due to SCC. In all but one case of failure due to pitting corrosion, SCC was found associated with the pitting but did not appear to contribute to the failure.
RETESTING
In 1988, retesting began on valve sections that had failed earlier as a result of SCC, either in-service or during the initial hydrostatic test. The retesting aimed to ensure the sections' continued integrity and reliability; they were retested to the same pressures to which they had been previously tested.
One valve section suffered a further SCC failure during the 1989 retest. Another valve section suffered one SCC-assisted failure during the 1988 retest, then subsequently suffered yet another SCC-assisted failure during a further retest in 1990.
Electron microscopy of the SCC rupture from the 1989 retest revealed that the flaw which led to the rupture and a 4.5-mm deep secondary crack adjacent to the rupture had propagated approximately 2 mm through the wall since 1986 . 2
Examinations of SCC less than 2 mm in depth could not determine whether the SCC had propagated, become dormant, or been initiated during the time between tests. It is apparent from these results that hydrostatic testing, although effective in removing near-critical flaws from the line, is incapable of permanently arresting the growth of all stress corrosion that is beyond a given threshold.
Fracture analysis of the stress-corrosion rupture that occurred during the 1989 retest suggested that the crack which led to the rupture had barely survived the 1986 test and was in fact close to theoretical critical size in service in 1989.
This analysis and results from the retesting program indicate that the minimum retesting frequency necessary to prevent in-service failures lies between 2 and 3 years for the highly susceptible valve sections.
EXCAVATIONS
More than 450 investigative excavations in association with the hydrostatic testing program allowed assessment of the prevalence and severity of SCC in the 914-mm line and permitted specific questions about SCC in TCPL's system to be addressed.
Fewer excavations were also conducted on a 508-mm (20-in.) line built in 1958, a 762-mm (30-in.) line built in 1958, and a 1,019-mm (40-in.) line built in 1982 to determine whether other lines in the system were also affected by stress corrosion.
The selection of excavation sites was based primarily upon environmental conditions. Consideration was also given to the results of aboveground surveys and the proximity of the line to residences and highways.
At each site, the line was excavated for approximately 50 in and bell holes were dug at three girth weld locations to provide access to the bottom of the pipe (Fig. 2). Samples from soil adjacent to the pipe and of ground water were collected from each site for analysis.
The condition of the coating was assessed visually and recorded. Where possible, samples of undercoating electrolyte were collected and placed in argon-filled, rubber-sealed vials for analysis. The pH of the undercoating electrolyte and a description of the corrosion deposits observed on the pipe surface were recorded for each patch of disbanded coating inspected.
The inspection patches were marked on the pipe to determine any correlation between electrolyte pH, corrosion deposit, and the occurrence of SCC. A high pressure (160 MPa; 23,000 psi) water jet was used to remove the coating and prepare the pipe surface of each inspection patch for wet fluorescent magnetic-particle inspection (MPI).
When a colony of SCC was detected with MPI, the following information was recorded: the circumferential location of the colony relative to the longitudinal seam weld, colony size, maximum and average crack length measured in the colony, minimum longitudinal distance between adjacent cracks, and maximum crack depth.
Crack depth was roughly estimated based upon the brightness of the MPI indication. If, however, the indication appeared particularly bright, the crack was sequentially ground until it had been removed and the remaining wall thickness measured with ultrasonics.
In instances for which the measured crack depth was more than 10% of the wall thickness, the pipe containing the crack was replaced. If the stress corrosion was less than 10% of the wall thickness, the pipe was recoated with field-applied polyvinylchloride tape coating and left in service.
EXCAVATION RESULTS
It became clear during the excavation program on the 914-mm line that polyethylene tape-coated pipe had a much higher propensity for SCC than did asphalt-coated pipe. SCC was detected at 69% of the tape-coated sites and only 14% of the asphalt-coated sites (Table 1).
The average number of SCC colonies detected at the tape-coated sites with SCC was almost twice the number detected at the asphalt-coated sites that had SCC. In addition, the depth of the SCC measured at the tape-coated sites was generally much deeper than at the asphalt-coated sites.
In total, almost 1,900 SCC colonies were detected on the 914-mm line during the investigative excavations. Only 1% of the colonies detected had a measured depth greater than 10% of the wall thickness, while more than 96% of them had a depth no greater than 5% of the wall thickness.
Therefore, although a vast number of colonies was found on the 914-mm line, the data indicate that in terms of safety and reliability, the line was in no immediate danger and that regular hydrostatic testing should be effective in controlling the problem by removing from the line those few colonies which grow to critical dimensions.
A method was developed for ranking the valve sections examined based upon the relative severity of stress corrosion in the section by considering the following parameters: maximum crack depth, maximum crack length, average colony size, and number of colonies detected.
The data collected to date indicate that the various valve sections exhibit widely varying degrees of SCC severity. Currently, eight valve sections are of considerably greater concern than the other valve sections examined.
The tape-coated sections exhibited a much higher degree of severity than did the asphalt sections with 16 of the top 17 valve sections in the severity ranking being tape coated.
In addition to the 914-mm line, SCC was also detected in TCPL's 508 mm and 762-mm lines and in the 1,067-mm line.
It is worth noting that the coal tar coated 508-mm line built in 1958 exhibited SCC with a maximum depth of 22% of the wall thickness, while the double wrapped tape-coated 1,067 mm line built in 1982 exhibited SCC with a maximum depth of only 3% of the wall thickness, thus illustrating the time dependency of SCC in the TCPL system.
SCC IN TCPL'S SYSTEM
Information obtained during the field program in conjunction with findings of parallel in situ environmental studies and numerous laboratory programs has helped to provide some understanding of the conditions under which SCC occurs in TCPL's system.
UNDERCOATING ELECTROLYTE
During the field program, 25 undercoating electrolytes, of sufficient volume for analysis, were collected from directly above stress-corrosion colonies.
Analysis of these electrolytes indicated that they were all very dilute, low pH (<7.5) solutions containing bicarbonate, carbonic acid, and several other ionic species (Table 2). These findings contrast sharply with the very concentrated, high pH (>9.5) solutions reported to be associated with the stress-corrosion failures experienced in the U.S. 3
The concentrated, high pH-type electrolyte was detected at some locations along TCPL's system, but there was no SCC or pitting associated with this electrolyte.
Initial attempts at generating stress-corrosion cracks in a simulated field environment proved unsuccessful until it was discovered that additions of 100% carbon dioxide to the simulated field environment resulted in the production of stress-corrosion cracks under extremely severe stressing conditions (that is, slow strain-rate tests). 5
In response to this significant discovery, a soils consultant was engaged to obtain data regarding the actual concentration of CO2 in the soil adjacent to the pipeline so that future growth-rate studies could be conducted under realistic conditions.
For this purpose, several soil gas and groundwater collection tubes were installed at three preselected sites in one of the more highly susceptible valve sections.
The study revealed that the environment adjacent to the pipeline is an extremely dynamic one with CO2 concentrations varying from a maximum of 23% in the spring to a minimum of 4% in the winter, depending upon topography and soil type. 6
It was also discovered that the groundwater adjacent to the pipe tends to become more acidic and concentrated with respect to the carbonate species during winter months compared with summer months (Table 3).
This study implies that, if the undercoating environment follows the same trends as the groundwater, the SCC in TCPL's system is likely growing only during the fall and winter months when the environment is most acidic and concentrated. It is worth noting that two of the SCC in-service failures occurred during the winter; the other occurred in the early fall.
BACTERIA IN SCC
Despite many published reports on the effects of anaerobic bacteria on general corrosion of buried pipelines, no information is available with respect to SCC.
Thus, to obtain some insight into whether bacteria may play a role in SCC, a few excavations were conducted in 1990 to collect samples of tape coating (bacteria being known to be able to live off the adhesive of such coating) for microbial analysis.
Subsequent analysis indicated the presence of both sulfate-reducing and acid-producing bacteria, in varying concentrations, on all of the tape samples analyzed. At SCC sites the concentration of the sulfate-reducing bacteria ranged from
Correlation of the microbial analysis with the severity of SCC detected tends to suggest that within a given area, the severity of the SCC (measured in terms of colony size, crack length, and crack depth) increases with increasing bacterial concentrations.
This observed correlation could result from the higher generation of such by-products as CO2, H2S, and organic acids (these perhaps assisting in the growth of SCC) that would be expected with the increased bacterial concentration.
Recent laboratory results indicate that additions of bacteria to the field-simulated environment result in reduced conditions. Such conditions appear to be a prerequisite for SCC in TCPL's system because the corrosion deposit most predominantly associated with SCC is white iron-carbonate which occurs under reduced conditions.
Thus, it is quite conceivable that the higher bacterial concentrations would result in a more reduced condition at the pipe surface and greater susceptibility to SCC.
CATHODIC PROTECTION
In addition to the collection and subsequent analysis of the undercoating electrolytes previously mentioned, more than 1,200 pH measurements were taken of the undercoating environments in order further to define the conditions under which stress corrosion occurs.
A total of 202 pH measurements were obtained from directly above stress-corrosion colonies.
The correlation of these pH measurements with the occurrence of stress corrosion conclusively illustrates that, regardless of the type of coating used, stress corrosion in TCPL's system is caused by an undercoating environment with pH
This correlation is significant because the low pH values indicate that the pipe surface at the area of a stress-corrosion colony is getting little, if any protection from the CP system and thus implies that stress corrosion occurs only in TCPL's systems when the pipe surface is inadequately protected by the CP system.
Investigations in the laboratory indicate that the pipe surface in the area of an SCC colony is receiving less than 100 mv cathodic polarization; at higher polarization levels, the pH of the electrolyte increases greater than 8.0 and consequently SCC will not occur.
Comparing Figs. 3a and 3b makes clear that it is substantially easier to protect cathodically the pipe surface, and thus prevent stress corrosion under asphalt coating, than under polyethylene tape. More than 40% of the pH readings taken in asphalt-coated sections were or equal to 8.5, compared with only 12% in tape-coated sections.
This observed difference in electrical current permeability between the two coating types is one of the most significant reasons that the occurrence of stress corrosion in asphalt-coated sections is so infrequent, compared to tape-coated sections.
SOIL,TOPOGRAPHY
In the tape-coated sections of TCPL's system, stress corrosion has been found to exist in all the various soils (muskeg, clay, silt, sand, and bedrock) and terrains present across the system. No apparent difference in the soil chemistry exists between stress corrosion and non-stress corrosion sites.
Stress corrosion is, however, predominantly (67%) located in imperfectly to poorly drained, high-resistivity soils (10 million ohm cm) in which reducing conditions are maintained.
During the 1990 held program, a hydrotest rupture occurred in the middle of a small creek. Examination of the pipe revealed that the severity of the SCC diminished rapidly towards the banks of the creek.
Preliminary results from an in situ environmental study of an area similar to the rupture site indicated that the oxygen level of 0.8% measured adjacent to the pipe beneath the creek is up to 10 times less than the oxygen level measured away from the creek. It is worth noting at this point that SCC was not detected in 10 excavations carried out in Saskatchewan in well drained, oxygenated soils even though the tape coating was severely deteriorated.
The major implication of these separate findings is that it may be possible to prioritize valve sections for remedial action based in part upon the oxygen permeability of the soil present in the section.
Topographically, SCC in tape-coated sections has been commonly located in the lower slope to level de.pressed areas where groundwater moves laterally towards and along the pipeline. SCC has rarely been located in stagnated areas. When detected in such areas, it was generally quite minor both in terms of prevalence and severity.
Most (83%) of the SCC detected in the asphalt-coated sections of TCPL's system was located in extremely dry terrains consisting either of sandy soils or a mixture of sand and bedrock. A possible explanation for this observation is that in these dry areas, it is extremely difficult to achieve the necessary level of uniform cathodic protection required to prevent SCC due to the low moisture content of the soil.
The remainder of the SCC was detected at sites that had localized areas of low pH environments and thus were probably inadequately protected by the CP system. It is worth noting that an in-service leak from SCC was detected in Manitoba during 1989. The leak occurred in an asphalt-coated section beneath a tape-mastic repair patch.
There was also a low-pH environment beneath the patch, indicating that it was shielding the pipe from CP currents.
COATED LINES
Although almost 70% of the excavation sites in tape-coated pipe contained SCC, it is important to understand that the condition of coating at most of these sites was generally, good.
In fact, SCC under polyethylene tape was rarely found in areas of badly disbonded coating or large holidays in the coating. The reason for this appears to be that these types of coating conditions enable the CP currents cathodically to polarize the pipe surface and consequently prevent SCC. Instead, the SCC is predominantly located close to longitudinal seam welds and girth welds where there is a minor amount of disbondment (less than 10 cm wide) as a result of the tenting effect of the tape coating that arises during its application.
The remainder of the SCC is found associated with areas of minor disbandment on the body of the pipe (that is, wrinkles) caused either by construction practices or soil stresses.
One main reason SCC is so rare in the asphalt-coated sections of TCPL's system is that the coating is generally in good condition with little disbandment. When SCC is detected, it is associated with areas of disbandment that are usually quite small and generally located in the top half of the pipe.
Fusion-bonded epoxy (FBE) coatings have been used for new pipe in TCPL's system since the early 1980s. FBE has not been found prone to disbandment when any of these lines have been exposed, and hence the necessary conditions for SCC noted above are unlikely.
Trans-Canada's current practice is to use only FBE where practical or other coating systems that are not prone to localized disbondment.
EFFECT OF TEMPERATURE, OPERATIONS
It quickly became apparent during the field program that the SCC present in TCPL's system does not display the same dependency upon line temperature as the SCC found in some U.S. pipelines and reported in the literature.
The industry's experience, as reported by Fessler, 1 has been that 90% of all in-service and hydrostatic-test failures attributable to SCC have occurred within 16 km downstream of a compressor station.
Fessler felt that this observed behavior was due to the higher temperatures and pressures associated with the discharge of the compressor station and that the effect of the higher temperatures probably predominates.
By comparison, TCPL's experience has been that only 50% of its SCC failures (service and hydrotest) have occurred within 16 km downstream of a compressor station, while more than 20% of them have been more than 30 km downstream of a compressor station with the farthest being 55 km downstream.
In addition to failure rates, temperature does not appear to affect the number of colonies detected or the depth of SCC that is less than 30% of the wall thickness (Fig. 4).
One possible explanation for the lack of a correlation between temperature and SCC in TCPL's system is that the solubility of CO2 in solution increases with decreasing temperature, thereby acidifying the solution and concentrating the carbonic acid species in the solution, both of which increase the probability of SCC occurring.
Thus, the effect of lower chemical activity associated with low temperatures may be offset somewhat by the increased corrosivity of the solution.
The four in-service SCC failures experienced by TCPL occurred at operating hoop stresses that were at least equivalent to 70% of SMYS of the pipeline steel.
However, there have been numerous findings of SCC in excess of 10% of the wall thickness at operating stresses much less than 70% of SMYS. In fact, one location, containing 85 stress-corrosion colonies with a maximum measured depth of 22% of the wall thickness, was operating at a stress level of just 50% of SMYS.
Based upon the obvious absence of a correlation between SCC and both line temperature and operating stress, TCPL will continue the program it initiated in 1990 to hydrostatically test second, third, and fourth valve sections which are up to 100 km downstream of compressor stations to ensure their continued integrity.
REFERENCES
- Fessler, R. R., "Stress Corrosion Cracking Temperature Effects," 6th Symposium on Line Pipe Research, Houston, November 1979.
- Christman, T. K., "Investigation of Hydrostatic Retest Failure, Confidential Report to TransCanada PipeLines," December 1989.
- Wenk, R. L., "Field Investigation of Stress Corrosion Cracking," 5th Symposium on Line Pipe Research, Houston, November 1974.
- Ogundele, G. I., "Stress Corrosion Cracking Susceptibility of Line Pipe Steels: Investigations on Crack Propagation," confidential report to TransCanada PipeLines, December 1987.
- Parkins, R. N., "Investigations Relating to the Environment Sensitive Fracture in the TransCanada PipeLines System," confidential report to TransCanada PipeLines, January, 1988.
- Marr, J. E., "Stress Corrosion Cracking-Carbon Dioxide Study at MLV 105," confidential report to TransCanada PipeLines, May 1990.
- Mackenzie, J. D., "Report on 1987 Pipe Integrity Program, SCC Research Program and Planned 1988 SCC Research Program," internal TransCanada PipeLines report, February, 1988.
- Justice, J. T., and Mackenzie, J. D., "Progress in the Control of Stress Corrosion in a 914 mm O.D. Gas Transmission Pipeline," NG-18/EPRG-Seventh Biennial Joint Technical Meeting on Line Pipe Research, Calgary, September 1988.
Copyright 1992 Oil & Gas Journal. All Rights Reserved.
