REHAB PERMITS DESERT LINE TO RUN AT ORIGINAL PRESSURES

A.M. Kurdi, M.S. Abougfeefa
Agip Oil Co. Ltd.
Tripoli

Alan K. Denney
John Brown Engineering & Constructors Ltd.
London

An extensive inspection and rehabilitation program on a 34-in. desert pipeline in gas-condensate service has restored the line to full operating pressures and ensured the line's active service life for at least 20 years.

Since Agip Oil Co. Ltd. built the 133-km pipeline in 1972 using API 5L Grade X60, it has suffered six known failures. There has been no single cause of the failures.

The four most recent failures have been recently reassessed. They relate to both internal phenomena (laminations/hydrogen induced cracking) and external phenomena (corrosion due to surface water and carbonate/bicarbonate stress corrosion cracking).

As a consequence of the failures, the line has been progressively down rated from the original operating pressure of 700 psig to 500 psig.

In order to upgrade the pipeline to an operating pressure of 600 psig, a rehabilitation program has been initiated based on data acquired from an intelligent pigging survey conducted in 1992, laboratory investigations, and data from previous refurbishment programs.

PRESSURE DE-RATING

One failure was reported on a weld during the initial hydrotest in 1972. A second, recorded in 1976, was a longitudinal crack occurring at kilometer point (KP) 23.5.

The line then operated at an inlet pressure of 700 psig until June 1984 when the first of three failures occurred. It resulted from localized external corrosion and occurred at KP 26.8 in a wadi area.

Following repair and during the hydrotest at 960 psig, a further failure occurred at KP 98.8 which resulted from a lamination defect.

After further repair, the line again failed under hydrotest at a pressure of 850 psig. The location was KP 58.8 and again the failure was associated with a lamination.

The line was then operated at 600 psig until November 1989 when failure occurred on a 48-m long section at KP 23.5. Following repair and hydrotest, the line was recommissioned at a reduced operating pressure of 550 psig which was further reduced in August 1991 to 500 psig.

These pressures resulted in operational difficulties relating to levels of condensation in the pipeline and the spasmodic nature of its flow.

Internally, there developed a corrosive fluid from the presence of CO2 dissolved in water forming carbonic acid. The line has been subject to corrosion-inhibition injection since November 1974.

The corrosion in the line has been monitored regularly by runs with instrumented in-line inspection tools ("intelligent pigs"). Three separate on-line intelligent pig surveys were carried out by one contractor between April 1984 and February 1989. The last survey on Jan. 17, 1992, was conducted by a different contractor.

All four surveys were carried out with magnetic flux leakage (MFL) intelligent pigs. These are generally reliable in detecting significant wall-thickness loss but lack the ability to detect stress corrosion cracking (SCC), hydrogen-induced cracking (HIC), delamination, and any form of longitudinal cracking.

Following each of the intelligent-pig surveys, a program of validation of the results was conducted using ultrasonics and cutting of suspect defects.

For each of the three earlier surveys, these showed significant inaccuracies in locations and over-reporting of the magnitude of the corrosion. The last survey, conducted in 1992, gave accurate measurements of pipe wall losses and locations.

Pipeline lengths showing significant magnetic anomalies (that is, suspect defects) were systematically examined, and a program of replacement of defective sections was carried out after all four surveys.

A detailed record was maintained of all defective areas in respect to type, location, and measured depth of corrosion.

OPTIONS

So that the line could again be operated between 650 and 700 psig, two options have been considered:

• Extensive study and investigation leading to inspection and replacement of suspect pipe.

• Total replacement of the line.

These options were evaluated from economic and safety points of view.

A major consideration was that the line will be operating in its current manner for only the next 2 years before changing to dry-gas transportation. The first option was therefore considered viable.

A strategy based on progressive refurbishment of the line was adopted:

• To take immediate actions to reduce risk of a failure in the next 2 years.

  This step involved elimination of all corroded areas exhibiting 20% or more reduction in thickness which will immediately permit an upgrade of operating pressure to 600 psig.

• To plan actions which must be taken for the line to operate for a further 20 years.

  This latter phase commenced immediately after the completion of the first action with the aim of further upgrading the line to 650-700 psig and allowing operation for the next 20 years.

UNDERSTANDING FAILURES

Before a rehabilitation program could be undertaken Agip considered it necessary to review and understand all the past failures and relate these to inspection and rehabilitation tasks.

The failure at KP 26.8 in 1984 was caused by deep external pitting resulting from the presence of surface water with a high chloride content combined with insufficient cathodic protection.

Following the failure, the segment of line in the wadi area was lifted above the level of the water table. Additional cathodic protection groundbeds were subsequently installed.

The gas in the pipeline contains approximately 4% CO2 by volume. In the presence of water this forms carbonic acid, which is corrosive to steel.

Theoretically there is potential within the pipeline for high levels of carbonic acid corrosion. Full modeling of the potential corrosion was done using the industry standard method of assessment.1-4

This takes into consideration the combination of temperature, pressure, and CO2 with modifying factors for the presence of inhibitor and, at higher temperatures, for the formation of a protective iron carbonate film.

Even when full allowance is made for reduction factors, high corrosion wastages are predictable. For example, at the inlet end of the line for a period of 19.6 years with a reduction factor to account for the use of inhibitor and a further factor for protective scale formation, corrosion of 24 mm is predicted.

Practically, this means that the 9.52-mm wall would have holed through and needed replacement several times in the life of the line.

From the intelligent pigging surveys, however, the defects due to carbonic acid corrosion have been less than 4 mm (Table 5 and Fig. 1) which is a fraction of the calculated possible corrosion values.

Hence, none of the failures to date has been caused by internal pitting reaching a point at which the remaining ligament could no longer withstand the pressure. This can be attributed to several factors, such as flow regime and effectiveness of inhibition treatments, as well as to protective adherent iron carbonate scale.

The model of CO2 corrosion shows a predicted peak of corrosion at about KP 20 at which point the temperature has decreased such that there would be no protective-scale formation.

Upstream of this point, corrosion is expected to be less because of protective-scale formation.

Downstream corrosion is expected to decline due to lower temperature and pressure.

A review of the location of severe defects detected in the surveys of 1984 to 1989, however, shows that they concentrate towards the two ends of the line (Fig. 1).

This pattern of observation of defects would be repeated if the less severe defects were included. This defies expectations from the theoretical model.

The pattern of circumferential location of corrosion also changes in that the more severe defects towards the inlet end tend to be between the 5 and 7 o'clock positions, whereas those at the downstream end show more sidewall defects.

One possible explanation is that sediment accumulation has reduced inhibitor effectiveness towards the downstream end of the line causing increased corrosion. The gas also contains an extremely small percentage of H2S, however, measured at between 5 and 10 ppm.

This level is well below that necessary to consider the conditions sour as defined by NACE MR01755 but laboratory examination of these sidewall defects shows them to be compatible with hydrogen-induced cracking (HIC).

HIC'S HISTORY

Hydrogen-induced cracking can develop in pipelines made of susceptible materials where there is a small content of H2S in the gas being transported. HIC is also known as "hydrogen pressure cracking," "blister cracking," and "step-wise cracking."

Its occurrence is relatively well documented within pipelines built from the controlled rolled steels of the 1960s and 1970s. The material of this line (Table 2) is typical of steels now known to be susceptible.

When the material was tested within this program of work using the industry standard test for susceptibility to HIC (NACE TM-02-846), it gave a figure of CLR of 96.7%, which indicates high susceptibility.

HIC occurs as a result of H2S in the pipeline gas keeping the hydrogen (forming as a result of corrosion) in the form of atomic hydrogen. The small size of atomic hydrogen permits it to migrate through the steel lattice where it combines to form hydrogen molecules at internal inclusion sites within the steel.

Controlled rolled steels provide many suitable sites, primarily Type 2 manganese sulfide inclusions. The molecular hydrogen has a large size which exerts a pressure on the lattice physically rupturing the steel lattice, forming stepwise cracks and blisters.

Significantly, scale samples removed from the line show the presence of both iron carbonate and iron sulfide (Table 3).

Detailed re-evaluation of the failures of July 11 and Feb. 12, 1984, shows that these were not caused by carbonic acid corrosion. Investigations attribute these two failures to HIC.78

A large blister measuring approximately 200 mm x 80 mm was found in the same pipe length as one of these failures. When examined, it showed delamination, one of the characteristic features of HIC (Fig. 2).

At the bending point of the blister a clear crack shows all the metallographic features of a sulfide stress corrosion (SSC) crack. This arises as a result of the very high local stresses occurring due to blister formation.

Another contributing factor, which was found during recent repair work to exist at the final portion of the pipeline, is sulfate reducing bacteria (SRB) considered to be introduced sometime during hydrotesting past activities.

1989 FAILURE

The most recent failure was at KP 23.5 (1989), near to which location a failure also occurred in 1976. No internal-corrosion features were observed in the failure area and the failure apparently originated from a colony of relatively short external cracks which joined up to form the failure (Fig. 3).

The cause was attributed to external carbonate-bicarbonate stress corrosion cracking.9 10 One of the characteristics of this type of cracking is that short cracks form and coalesce with adjacent ones to reach the critical size for fast fracture.10

There are two different forms of SSC in high-pressure gas lines from the soil side, the commonest giving intergranular cracks and the other giving transgranular cracks.

These correspond, respectively, to high pH and low pH solutions being generated and causing the cracking. The high pH form also relates to higher operating temperatures.

The different features of these two forms have been listed by Parkins (Table 4).11 The cracks were found to be intergranular and to have a marked 45 orientation, characteristic of high pH carbonate-bicarbonate stress corrosion cracking with this type of steel (Fig. 4).

There is a strong correlation between the incidence of intergranular SSC and temperature on pipelines, the incidence decreasing with decrease in temperature. For example, in moving along a line from gas-inlet points or compressor stations, the risk of occurrence declines.12

Moreover, this type of cracking only occurs within a relatively narrow range of potential at the steel surface. The corrosion originates from the presence of a solution in which hydroxyl ions are generated by the potential arising from the CP system.

This facilitates the pick-up Of CO2, giving rise to the carbonate-bicarbonate solution, which becomes concentrated by CP effects and by the heat from the line.

SURVEY RESULTS

An on-line survey was carried out in January 1992 in parallel with the studies previously summarized. This enabled Agip to achieve a complete picture of the current condition of the 34-in. diameter gas pipeline.

The log submitted classified all pipe lengths according to the worst magnetic anomaly (broadly speaking, the maximum depth of volumetric corrosion) detected within the length.

The log showed the following graded pipe lengths:

• 2,019 Grade 1 (light) anomalies, representing a wall loss of 15-30%, approximately

• 123 Grade 2 (moderate) anomalies, representing a wall loss of 30-30%, approximately

• A single Grade 3 (severe) anomaly, representing a wall loss of more than 50%.

After the completion of the intelligent pigging survey, verification digs were conducted and residual thicknesses determined with ultrasonics. Three of the moderate anomalies were inspected, confirmed (with gamma-ray and ultrasonics), and cut.

Table 5 shows the outcome of the verification digs. Fig. 5 shows one such area as surveyed.

Following receipt of the on-line survey report, a predetermined number of suspected defects (at varying distances, orientation, and amplitude along the entire pipeline) were selected for excavation and inspected as confirmation of the integrity of the final submitted data.

Table 6 presents the results of these confirmation digs.

ACCEPTABLE PRESSURES

Part of the engineering work on the pipeline consisted of a detailed study of the acceptable operating pressures for the line. Conventional calculations of the safe operating pressure of a corroded pipeline utilized ASME/ANSI B31.8 Appendix L.

Several different operating pressures were considered and related to different remaining wall thicknesses and the different defects detected by the 1992 intelligent-pig survey (Table 7).

This analysis suggested that the removal of all defects causing a wall reduction of more than 30% corresponded to an operating pressure of 600 psig.

This would be acceptable, however, if the only source of defects was carbonic acid (sweet) corrosion. The analysis does not evaluate risks due to planar defects such as HIC or SCC.

To consider this aspect, a study was made of the toughness of the linepipe materials. This aspect was correlated to critical crack sizes for bursting at different operating and hydrostatic test pressures, resistance of the linepipe to a running fracture, and detectability of planar defects by intelligent pigging.

The major concern was that the failures to date had largely been caused by crack-like defects (HIC and SCC) which are difficult to detect by any available intelligent pig. They are hence difficult to eliminate from the line.

Mechanical tests conducted included tensile tests, Charpy V-notch impact test, CTOD, and KI-SCC tests.

The CTOD tests determine the toughness of the linepipe; these results were analyzed to give crack sizes which would be critical for different peak pressures of the line (Table 8).

The results of the KL-SCC tests can be used to establish a size of defect which could grow in the presence of H2S, without increase in stress, with the likelihood of pipeline rupture occurring when the sizes listed in Table 8 are exceeded.

Where lines have been found to suffer external cracking, it is an acknowledged technique to hydrotest them repeatedly to failure in order to burst defective areas. This eliminates critical defects.

Additionally, this method results from the lack of proven intelligent pig nondestructive testing (NDT) methods for detection of cracks.

The method of analysis given here can be used to calculate a suitable hydrostatic testing pressure which will eliminate not only crack-like defects already present but also smaller defects which could otherwise grow under the influence of H2S to a critical size for failure.

REHABILITATION PROGRAM

Once the pipeline condition had been determined and a reasonably accurate picture obtained of the possible location of the defects along the entire 133 km pipeline length, a rehabilitation program was drawn up. This was divided into immediate and long term phases.

On Oct. 15, 1992, the rehabilitation of the 34-in. gas pipeline started. The program consisted of the elimination of the 120 anomalies, of which one was severe. In the first 40 km of the pipeline, 85% of the defects were accommodated, with the majority of the remaining 15% positioned at and around KP 120.

The first two cuts of the pipeline were selected at the lowest two points in the first portion of the pipeline to drain from the system as much water as possible.

The repair program was set to work in a chain-like manner, that is, excavation, followed by wrapping removal, inspection (ultrasonic thickness gauging and, where possible, gamma ray), cutting, welding, 100% gamma-ray inspection, magnetic-particle inspection, rewrapping, and finally holiday detection prior to back filling.

A problem arose during the welding of the new pipe to the original. The problem was caused by residual magnetism arising from the powerful magnetic field induced by the intelligent pig instrument (Fig. 6).

This was overcome by wrapping electrical cables around the two pipes to be welded (Fig. 7), placing the ground electrode at the welding gap, and gradually increasing the current from 0 until normal welding was achieved.

As a result of both the verification and confirmation digs, all the selected 120 anomalies were detected and replaced.

Because of the constraint of material availability on site, however, areas of immediate concern were replaced: those of 20% reduction in thickness or more (Table 10), rather than replacing complete joints or immediate joints showing light corrosion, which otherwise would have been replaced.

It was also clear from visual observations made on cut-out pieces that a substantial number of the these defects are located at or near welds. This suggests the defects may be attributed to bad welding or the effects of the heat-affected zone.

To ensure the safety of important foreign pipeline crossings, crack stoppers were installed 3 m either side of each crossing.

Prior to the installation, a distance of 200 m either side of each crossing was inspected visually and internally from on-line survey logs and as much as possible by magnetic particle inspection.

This was done to safeguard these lines against any unforeseen failure of the 34-in. line.

One finding of this rehabilitation work was that in the first part of the pipeline, especially in the first 80 km, faults are mainly dominated by CO2 (sweet) corrosion located between the 5 and 7 o'clock positions producing, in places, a pronounced groove. Between KPs 100 and 120, corrosion is very evident which occurs anywhere around the pipe circumference with no specific orientation.

The rehabilitation program took only 7 weeks to complete rather than the 10 originally scheduled. The pipeline was then hydrotested at 750 psig to operate at a pressure of 600 psig. This will temporarily satisfy the need of the consumer.

Agip is currently planning a long-term program to upgrade the operational pressure of the pipeline to 700 psig and extend its life span for a further 20 years, as an alternative to replacing the entire line.

This program, anticipated to take 2 years, includes improvement of the existing cathodic-protection system, especially in the first 40 km, following potential and coating attenuation surveys, and improvement of monitoring systems.

The plan also features ultrasonic intelligent pigging using a tool customized to detect HIC, SCC, and laminations. Comparison will also be made with the 1992 intelligent pig survey, but wall losses are seen as the less critical defects.

Preparation for this survey will involve examination of samples of linepipe by an on-line survey contractor for trial and calibration purposes. This will use pipe lengths already removed from the line.

REFERENCES

1. de Waard, C., and Milliams, D.E., "Prediction of Carbonic Acid Corrosion in Natural Gas Pipelines," Industrial Finishing and Surface Coating, November 1976, pp. 24-28.

2. Smith, L.M., and de Waard, C. "Selection Criteria for Materials in Oil and Gas Processing Plants," U.K. Corrosion '87, Brighton, October 1987.

3. Simon-Thomas, M.J.J., de Waard, C., and Smith, L.M., "Controlling Factors in the Rate of CO2 Corrosion," U.K. Corrosion '87, Brighton, October 1987.

4. Smith, L.M., and de Waard, C., "Material Selection for Oil and Gas Processing Plants," Industrial Corrosion, December 1989/January 1990, pp. 14-18.

5. NACE Standard MR0175-92, "Sulfide Stress Corrosion Cracking Resistant-Metallic Materials for Oilfield Equipment," National Association of Corrosion Engineers.

6. NACE Standard TM-02-84 Test method, "Evaluation of Pipeline Steels for Resistance to Stepwise Cracking."

7. Doughty, A.S., "Metallurgical Examination of Sections from 34-in. Diameter Gas/Condensate Pipeline," Capcis, April 1992.

8. Denney, A.K., "Engineering Investigation and Assessment of 34-in. Pipeline, JBE&C, May 1992.

9. Booth, G.S., and Threadgill, P.L. "Investigation into the Failure of a 34 in. Diameter Gas Line," the Welding Institute, October 1990.

10. Parkins, R.N., "Pipeline Failure Investigation," April 1992, University of Newcastle-upon-Tyne, England.

11. Parkins, R.N., "Topical Report on Environment Sensitive Cracking (low pH stress corrosion cracking) of High Pressure Pipeline," American Gas Association, NG-18 Report No. 191, August 1990.

12. Fessler, P.R., "Stress Corrosion Cracking Temperature Effect," 6th Symposium on Line Pipe Research, 1979, American Gas Association.

Copyright 1993 Oil & Gas Journal. All Rights Reserved.