K.E.W. Coulson, R.G. Worthingham
NOVA Corp. of Alberta
Calgary
Guidelines for more accurate assessment of pipe corrosion-damage severity have been developed as a result of research by NOVA Corp. of Alberta.
Following these damage-assessment guidelines could help pipeline operators prevent unnecessary and expensive pipe repair or removals.
This first part of a two-part series on this research presents the background and establishes the kind of corrosion under scrutiny.
The conclusion presents the results of burst tests performed by NOVA.
These guidelines resulting from NOVA's research have been successfully used by NOVA to analyze data from in-line inspection of the 20-in. Peace River main line.
IN-SERVICE FAILURES
Analysis of pipeline failure statistics from the U.S. National Transportation Safety Board has identified the two main causes of in-service failures on high-pressure gas-transmission pipelines. The causes are mechanical damage and corrosion.1
To ensure the continued safe, reliable operation of a corroded pipeline, the operator will assess the influence of this type of damage by hydrostatic testing or applying industry-accepted, corrosion-assessment criteria, such as those found in ANSI/ASME B31G or CSA Z184.2 3
The economic disadvantages of hydrostatic testing of corroded pipelines to ensure fitness for purpose have been previously documented.4
Although more economic than hydrostatic testing, the currently accepted industry criteria methods are still overly conservative. Following these conservative approaches, pipeline operators could remove pipe containing noninjurious corrosion damage which has been mistakenly assessed as unacceptable.
NOVA Corp. of Alberta undertook research to understand better the failure behavior of corrosion defects and develop a basis for the more accurate (less conservative) assessment of their significance.
Some corrosion-damage geometries which are not specifically addressed by the currently accepted industry criteria were also studied.
"INJURIOUS" CORROSION
The economic implications of pipeline maintenance first became significant in the 1960s when maintenance costs increased dramatically because of corrosion-induced pipeline failures. At this time, industry recognized that corrosion damage to pipelines had to be considered during routine maintenance.
Pipeline codes were modified to require that "injurious" corrosion damage be repaired or replaced.5 Unfortunately, no method for identifying what damage was injurious was provided.
As a result of the need for defining injurious corrosion damage, work was sponsored by a major U.S. gas-transmission company and the Pipeline Research Committee of the American Gas Association (AGA). This work produced the NG-18 surface-flaw equations in the early 1970s.
These calculations were complex and laborious to perform longhand because slide rules and log tables were the main computational tools available at the time they were done.
To make these equations more useful, certain assumptions and simplifications were made. The simplified equations have since been incorporated into the ANSI/ASME B31G2 and CSA Z1843 codes as a method for the evaluation of injurious corrosion damage.
The simplified equations used in the pipeline codes deliberately contained large safety factors (at least 53% conservative for 0.72 design factor and location Class 1) to prevent pipeline failures. The assumptions and limitations of the simplifications included:
- Parabolic flaw shape, determined by maximum depth and projected axial length only
- Simple corrosion configurations considered
- Steel flow strength conservatively set at 111% specified minimum yield strength (SMYS)
- Design factor set at 0.72 only (this is location Class 1)
- Other location classes not recognized
- Use of overdesigned and heavy-wall pipe sections not recognized
- Flaws with depths in excess of 80% of the wall thickness (W.T.) not acceptable
- No differentiation between leak or rupture failure condition. Inclusion of leak data in verification burst tests resulted in additional conservatism in the equations.
Use of the simplified code equations did, however, provide the much needed definition of injurious corrosion damage.
A pipeline operator could apply these simplified code equations and be reasonably confident that failures would not occur because of unsafe corrosion damage remaining in pipelines and be spared the expense of performing unnecessary repairs to superficial corrosion damage.
Experience with the simplified equations and analysis of their limitations, however, revealed that acceptable defects were being identified by the codes as requiring repair.
With the advent of reasonably priced computers and programmable scientific calculators, it became feasible to use the original NG-18 surface-flaw equations and perform analyses free of many of the limiting assumptions contained in the simplified equations.
Both ANSI/ASME B31G and CSA Z184 permit the performance of "an engineering assessment" or a "more rigorous fracture mechanics analysis" without specifying what method or methods may be used. These clauses allow use of the original NG18 equations.
CORROSION DAMAGE
The NG-18 surface-flaw equations assess the significance of corrosion damage on pipelines with respect to rupture failures .6 Flaws with depths in excess of 80% W.T. require repair to prevent leak failures.
The basic form of the NG18 surface-flaw equation is given in Equation 1 (accompanying box; also see Fig. 1). Equation 2 gives the form for a three-term Folias factor; Equation 3, a two-term Folias factor.
Equation 1 may be rearranged to solve for failure pressure (FP) (Equations 4, 5, and 6) or maximum allowable pit depth (Equations 7 and 8). Equation 4 is the general form used in determining expected failure pressure.
The actual corrosion shape may be approximated by either the parabolic or rectangular models.
If the corrosion is rounded in shape (Fig. 2), the metal loss may be approximated with the parabolic model (Equation 5). This model assumes the corrosion cross-section to be parabolic with a maximum depth equal to the measured maximum pit depth (D).
A = 2/3 DL
For the parabolic model, maximum pit depth (D) and axial length (L) are the corrosion measurements required.
If the corrosion is irregular in shape (Fig. 3a), the parabolic model will either overestimate or underestimate the actual amount of metal loss. In this case, the rectangular model (Equation 6) should be used to obtain a more accurate assessment.
This model assumes the corrosion cross-section to be a rectangular notch (Fig. 3b), with depth equal to the average depth of corrosion (C)).
A = DL
The average depth must be determined by obtaining sufficient depth measurements inside the corroded region being assessed to provide an accurate representation of the actual amount of metal loss.
Depth measurements at 5-mm intervals along the corrosion length have been found satisfactory for this purpose. For the rectangular model, average depth (D) and axial length (L) are the corrosion measurements required.
Calculation of maximum allowable pit depth provides a sensitivity check on the amount of additional metal loss which would be required before a failure could occur.
The reasons for using the parabolic and rectangular models in these calculations are the same as previously described.
ACCEPTANCE CRITERIA
The objective in performing an assessment of corrosion damage is to ensure at least the same level of integrity as would be obtained from a hydrostatic test of the pipeline in question.
CSA Z184 specifies minimum hydrostatic test pressures for the various location classes of pipelines.
The minimum hydrostatic test pressure for Classes 1 and 2 locations is 125% of the maximum allowable operating pressure (MAOP). For Classes 3 and 4, and meter stations, the minimum hydrostatic test pressure is 140% MAOP.
In Alberta, compressor stations are regulated by the Alberta Boilers Branch and must comply with ASME Section VIII UG-99. The standard hydrostatic test pressure in compressor stations is 150% MAOP.
Flaws with expected failure pressures less than the requirements stated require repair.
As well, flaws with maximum or average depths deeper than the maximums allowable as calculated by Equations 7 and 8, based on the requirements stated for minimum hydrostatic test pressure, require repair.
The following data are required before an assessment of corrosion damage should be initiated:
Pipe: Diameter (nominal)
W.T. (nominal or actual, if available)
Flow stress (pipe grade + 68 950 kPa)
MAOP
Location class.
Corrosion: Maximum depth or average depth, depending upon model used.
Length projected on longitudinal axis.
BEYOND PRESENT METHODS
During corrosion investigation excavations on NOVA's 20-in. Peace River main line and the 36-in. Western Alberta system extension in the early 1980s, it became apparent that certain corrosion geometries were beyond the scope of the damage-assessment methods available.
It was decided to develop methods for assessing these types of corrosion geometries before the British Gas in-line inspection data analysis.
A research project was initiated to investigate the significance of spirally oriented corrosion, corrosion longer than 1D (1 pipe diameter), and interacting or adjacent corrosion defects.
- Spiral corrosion typically occurs on tape-coated pipelines.
The problem results primarily when the field-applied tape-application machine advances too fast.
The tape is wrapped on the pipe with no overlap, leaving bare steel exposed between the tape spirals (Fig. 4). Frequently, a nonadhering fiber glass-reinforced asphalt, Kraft paper, or thin polyethylene outer wrap was applied over the single layer of polyolefin tape.
Metal loss, in a spiral pattern, primarily occurred on NOVA's tape-coated pipelines, before application of cathodic protection during the early years of operation.
Typically, corrosion of the exposed bare steel was arrested with cathodic-protection application.
Ongoing corrosion has been observed to occur, however, where the cathodic protection is shielded by the nonadhering outer wrap.7 The resulting defects were typically oriented 20 from the pipe circumference, with a maximum observed angle of 45.
The NG-18 equations were not developed to calculate the expected failure pressure of the spiral defects which result from a corrosion process becoming active on these exposed or shielded bare-steel areas. The equations require several parameters as input, one of which is the axial projected length of the corroded area.
Elementary mechanics analysis of the situation suggested the use of axial projected length was not appropriate in assessing spiral corrosion damage.
- Long corrosion. Axially oriented wrinkles in plastic tape coatings are formed by pipe or soil motion. Groundwater penetrates beneath the wrinkles but, unfortunately, cathodic-protection currents are shielded from the exposed pipe under the wrinkle, resulting in axial corrosion defects.
Because of the nature of these wrinkles, there is no practical limit to their length and the length of corrosion beneath them. Flaws longer than 1D have frequently been observed on NOVA's tape-coated pipelines.7 In the NG-18 equations, the effect of defect length is addressed in the "shape" or Folias-factor equation.
The three-term Folias factor most often used with the NG-18 surface-flaw equations realistically describes the effect of length on failure pressure for lengths less than 1D (Equation 2).
For longer defect lengths, this factor provides erroneously high estimates of failure pressure.
A two-term Folias factor is available for these long defect lengths but is more conservative than the three-term factor (Equation 3). The behavior of these defects required investigation to permit the accurate assessment of damage severity.
- Adjacent defects. The NG-18 equations were developed for the assessment of isolated defects. In reality, however, corrosion defects often occur in groups or clusters.
It is obvious that, as the distance between two defects decreases, they will, at some unknown separation, begin to behave as a single, more severe defect. Knowledge of this critical separation distance is extremely important when the significance of adjacent corrosion defects is being assessed.
Early attempts to investigate the effect of defect interaction with the NG-18 equations have proven unsuccessful .6 The defect sizes and separations used were such that no interaction occurred and the defects behaved as individuals.
ANSI/ASME B31G 2 addressed "isolated" corrosion pits only and provided no guidance for dealing with adjacent defects.
The initial approach used by NOVA assumed full circumferential interaction occurred. That is, a reduction in pipe strength occurred regardless of the position of defects because the projected axial length of all defects was considered the defect length (Fig. 5).
This approach was quite conservative but provided a means for assessing adjacent corrosion damage when the codes provided, no guidance. CSA adopted one defect length or width (depending on the corrosion configurations) as the critical separation distance in Clause 10.10.6.3 This was obtained from fracture-mechanics approaches used in Z184 for assessing girth weld defects.8
British Gas, on the other hand, has developed a four-pipe wall thickness (4 x W.T.) criterion as its critical separation distance. Discussion with British Gas indicated investigations performed on pipes containing stress-corrosion cracks revealed these critical separation distances.9
The debate within the industry on this topic has made the investigation of adjacent defects necessary.
REFERENCES
- Eiber, R.J., Jones, D.J.,and Kramer, G.S., "Outside force causes most natural gas pipeline failures, " OGJ, Mar. 16, 1987, p. 62.
- ANSI/ASME B31G-1984. Manual For Determining the Remaining Strength of Corroded Pipelines, ASME, New York.
- CAN/CSA-Z184-M86, 'Gas Pipeline Systems," Canadian Standards Association, 178 Rexdale Blvd., Rexdale, Ont., September 1986.
- Shannon, R.W.E., and Argent, C.J., "Maintenance strategy set by cost effectiveness," OGJ, Feb. 6, 1989, p. 41.
- CSA Standard Z184-1968. Gas Transmission and Distribution Piping Systems, Canadian Standards Association. Mar. 18, 1968.
- Keifer, J.F., "Fracture Initiation," 4th AGA Symposium On Line Pipe Research. Battelle Laboratories, Columbus, Ohio, Nov. 18-19, 1969.
- Worthingham, R.G,, "On-line Inspection, Analysis and Maintenance of the 914 mm Western Alberta System Extension," November 1985, NOVA Technical Report #19131.
- CAN/CSA-Zl84-M86, Gas Pipeline Systems, Appendix K, "Standards of Acceptability for Circumferential Pipe Butt Welds Based on Fracture Mechanics Principles."
- Jones, D.G., British Gas Corporation, private correspondence, July 22, 1986.
- Maxey, W.A., "Y/T Significance in Line Pipe," 7th AGA Symposium on Line Pipe Research, Houston, Oct. 14-16, 1986.
Copyright 1990 Oil & Gas Journal. All Rights Reserved.
