PRODUCTS PIPELINE REHABILITATED WHILE ON STREAM

Jan. 9, 1995
A.K. Denney, S.L. Coleman, R. Pirani John Brown Engineers & Constructors Ltd. London N. Webb Corrolec and Metallurgical Services Rivonia, Republic of South Africa P. Turner Teknica (Overseas) Ltd. London Rehabilitation of a 186-mile petroleum products pipeline in southern Africa employed sleeve welding, reinstatement of external coatings, and upgrading of the cathodic-protection system. The pipeline had an unusual history in which the political environment of the region forced its shutdown for
A.K. Denney, S.L. Coleman, R. Pirani
John Brown Engineers & Constructors Ltd.
London
N. Webb
Corrolec and Metallurgical Services
Rivonia, Republic of South Africa
P. Turner
Teknica (Overseas) Ltd.
London

Rehabilitation of a 186-mile petroleum products pipeline in southern Africa employed sleeve welding, reinstatement of external coatings, and upgrading of the cathodic-protection system.

The pipeline had an unusual history in which the political environment of the region forced its shutdown for 17 years. This shutdown played a major role in its deterioration.

The pipeline, which exhibited extensive internal and external corrosion, was a crucial supply route for imported refined products. So important was the line that during the entire repair project, the line could not be shutdown.

This technical difficulty was compounded by various practical difficulties as well.

SHUTDOWN

The 300-km 10.75 in. OD x 0.250-in. W.T. pipeline was fabricated from API 5L X-46 linepipe. It was originally installed in 1964 and commissioned in 1965 to transport crude oil to a (then) new refinery.

Its main facilities are two pump stations and one terminal.

The shutdown of the pipeline lasted from from 1965 to 1982. During shutdown, maintenance of the cathodic-protection (CP) system was impossible. This caused most of the external corrosion.

Crude oil was also shut in during this period. In 1981, this crude was evacuated from the line and the line converted to multi-products because the refinery downstream had fallen into disrepair.

Refurbishment began in 1981 with replacement of 40 km of the most heavily corroded pipe (Fig. 1) , and another 40 km lifted, cleaned, and rewrapped. The line was recommissioned and operations begun in 1982.

In 1983, a close-interval potential survey was followed by a program of wrapping repair and replacement.

In 1987, to improve the operating condition of the pipeline, the pipeline operating company initiated further investigations of the status of the pipeline. This effort led to the work under discussion here.

COATINGS, CP HISTORY

The original wrapping system was a fiber glass reinforced coal-tar enamel; the impressed current CP used 14 transformer rectifier sets with horizontal groundbeds consisting of silicon iron anodes in coke breeze backfill.

As part of the 1981 refurbishment, the 40 km of replacement pipe was coated with factory-applied extruded epoxy/polyethylene coated pipe.

The lifted length was stripped and rewrapped with a polyethylene/butyl laminate tape. Further localized wrapping repairs were carried out with Denso petrolatum innerwrap with an adhesive polyethylene outerwrap.

At the same time, the CP system was re-designed and new groundbeds and transformer rectifiers (TR) installed. Some of the old TR sites were abandoned, and additional units installed where insufficient spread was obtained.

New power supplies and standby diesel alternators were also installed.

There was no assessment of internal corrosion during this early work.

Once the pipeline was operational, an overline close-interval polarized potential survey was conducted to determine the effectiveness of the CP system and locate any areas which still required wrapping repair.

One of the most significant results of the survey was the accurate status record of the first 18 km of the pipeline. This was one of the most corrosive areas of the route, being rice fields, and also had the greatest potential for consequential damage should a leak occur.

Consultants had proposed that this section of the line be replaced, but the close-interval survey showed that only 1 km of pipe required rewrapping.

Excavation revealed that this section had no corrosion, despite extensive loss of wrapping, because it was adjacent a CP station. The result was a major cost savings.

A priority-based repair program was instituted to carry out the necessary repairs to the wrapping system. These repairs were effected with the (then) relatively new polyethylene-backed rubberized bitumen tapes.

A local product was used because it was found that the U.K.-sourced tapes tended to flow at high ambient temperatures. This tendency presented a storage rather than a service problem.

One of the difficult aspects of this phase of the program was the existence of significant lengths of pipeline on which the CP potentials were inadequate, but there were no suitable groundbed sites and power supplies were few and far between.

These areas tended to have relatively high resistivity with low resistivity hot-spots to the extent that made achieving protection by increasing the output of the TRs impossible.

Extensive economic analysis of the alternatives and a view toward possible future upgrading led to a decision to use magnesium ribbon as an interim measure to provide protection to these areas. The ribbon was designed to provide a 10-year life.

As part of the rehabilitation work carried out by the authors' companies, starting in 1987 a further overline potential survey was conducted, as well as an intelligent pig survey.

The results of these surveys were correlated to provide a definitive schedule of necessary repairs, based on both metal loss indications and wrapping/cathodic protection deficiencies.

PIPELINE INSPECTION

Pipeline intelligent pigging was carried out between September 1989 and January 1990 with a British Gas On-Line Inspection vehicle.

Six runs were carried out, consisting of one pass with a multigauge vehicle and five with the inspection vehicle. The report confirmed that the pipeline had suffered extensive internal and external corrosion.

The internal corrosion was predominantly located in the 4-8 o'clock quadrant and, in some sections of the pipeline, appeared to extend throughout the bottom section. The external corrosion was more randomly distributed in all quadrants and along the entire length of the pipeline.

An automatic computerized process was used to grade the metal-loss features according to their estimated depths. This information was displayed as a list of sequential girth welds, with metal-loss indications identified by their distances from the upstream girth weld and by their orientations viewed in the direction of flow.

A code system was utilized to identify the metal loss as follows: no star = 030% metal loss; * = 30-50% metal loss; ** = 50-70%; and *** = 70% and higher.

Distribution of the metal-loss indications along the pipeline was shown on histograms based on the number of features per 100 m of line length. (An example is shown in Fig. 2 (10341 bytes).)

No original chainage records were available and the line has an erratic path in places. This resulted in some practical problems in relating the location data from the intelligent-pig survey to the pipeline as laid.

The line had been constructed with double random lengths of pipe, and the variable lengths of these were used to advantage to assist with locating defects. Where repairable events were detected and not readily found, the pipeline was excavated for at least three pipe joints.

These were measured and compared with the circumferential weld indications on the pig data log and used to give a precise local fit to the repairable event. It was also found to be necessary to confirm reported events with ultrasonics.

Re-evaluation of the intelligent pig data was requested from the pig operator when the reported "serious events" could not be located. In several such cases this more detailed analysis resulted in the significance of the defect being reduced.

REPAIR PROGRAM

The results of the intelligent pigging were used to establish a repair program that gave priority to the more corroded areas (three-star events, or metal loss of 70% or more).

The extent of corrosion identified by the intelligent pigging led to approximately 40 km of pipeline being replaced with new API 5L X-46 line pipe.

Hot tapping was used to connect and commission the new sections of pipeline with minimum disruption to the pipeline operation.

It is interesting to note that some of the sleeving was required on sections which had been rewrapped during the first reinstatement program. These pipes had severe internal corrosion from the years of standing stagnant crude oil.

External sleeves were used on all the three-star events. The opportunity was taken also to include any two-star events (50-70% metal loss) in the immediate vicinity of the three-star events.

SLEEVING ON A LIVE LINE

Welding external sleeves onto a live pipeline, a delicate process under any circumstances, was made more complex on this project because sleeves were being attached to areas in which internal and/or external corrosion had caused significant loss of wall thickness.

The problem was therefore to derive a suitable sleeving and welding procedure which would not cause "burnthrough" of the pipe resulting in the sudden catastrophic release of the pressurized fluid '

Consideration was given as to whether to use half or full sleeves to repair the line.

Half sleeves would have been attached to the line with both longitudinal and circumferential fillet welds; full sleeves can be attached to the pipe with a circumferential fillet weld only. (The longitudinal weld for full sleeves is sleeve-to-sleeve only and not sleeve-to-pipe.)

Internal pipeline pressure significantly influences pipe wall failure: a longitudinal weld failure occurs at a much lower pressure than for a circumferential weld.

For that reason, full sleeves were selected for the pipeline refurbishment. The configuration of the sleeves is shown in Fig. 3 (8909 bytes).

Pipe-wall thicknesses on the rest of this pipeline were so reduced and the risk of "burnthrough" during sleeving was so high that a study was commissioned from specialist consultants to determine safe conditions for welding.

A personal computer helped calculate the maximum pipeline-surface temperatures during welding under different conditions. The program was based on an experimentally verified model.

It calculated temperatures vs. time for various locations near the weld bead as the weld is deposited. Heat transfer to the pressurized, flowing products inside the pipeline was accounted for by considering the relevant properties of the fluid (e.g., specific heat, density, surface tension, and latent heat) and its pressure and flow velocity.

The output from the computer runs fell within surface temperatures for the various wall thicknesses, heat inputs, and fluid flow velocities. The result for the worst case, a 0.160-in. W.T. pipe, is shown in Fig. 4 (7153 bytes).

Fig. 4 (7153 bytes) shows that the highest predicted inside surface temperature for a heat input of 25 kJoules/in. is 1,685 F. (919 C.). This is below the 1,800 F. (982 C.) level which is considered a threshold above which burnthrough is risked. All other conditions produce lower inside surface temperatures.

On site welding was only carried out on pipe with a remaining wall thickness (from ultrasonic determinations) greater than 0.250 in. and when diesel was flowing (as opposed to Avgas or gasoline).

The calculations showed that this was a safe procedure with no risk of burn through.

The study also considered the possibility of blowout.

During welding, the yield strength of the weld pool and heat-affected zone (HAZ) are reduced to such an extent that they cannot carry any load.

Blowout occurs when the remaining wall thickness cannot contain the internal pressure.

The critical internal pressure above which blowout would occur was calculated for welding to a 0.160-in. W.T. pipe with 15 kJ/in. welding heat input and was found to be 53 bar, well above the 34-bar operating pressure.

Experimental data showed that blowout was preceded by internal-surface temperatures greater than 1,800 F. (982 C.), the critical temperature for burn through.

This indicated that the theoretical model was conservative and that there was no significant risk of blowout during welding of 0.16-in. thick material at an operating pressure of 34 bar with the shielded metal-arc welding (SMAW) procedures adopted.

DEVELOPING A PROCEDURE

Trying to avoid burn through necessitates minimizing welding-heat input. This can result in rapid cooling of the heat affected zone and can create the problem of hydrogen cracking there.

Hydrogen cracking is normally avoided by application of preheat where heat inputs are low and by controlling the moisture content of the welding consumables. Preheating of a fluid-filled pipeline is not normally feasible especially if the liquid is flowing.

Weldments made with the gas metal-arc welding (GMAW) and GTAW processes are less susceptible than others to hydrogen cracking. These processes, however, are unsuitable for pipeline welding in the field.

The solution was to use a low hydrogen (E7018) SMAW consumable suitably stored, baked, and held in a heated quiver prior to welding.

The following welding and sleeving procedure was developed based on the safety recommendations and was incorporated into the sleeving specification produced for the work:

  1. Remove coating and oxide by mechanical wire brushing.

    Establish by visual or ultrasonic examination the precise position of the thinned area of the pipe that had been detected by the intelligent pig. The sleeve was positioned to ensure this area lay midway along the sleeve.

  2. Carry out ultrasonic thickness check around the full circumference at the proposed location of the sleeve-to-pipeline circumferential fillet weld.

    Establish the minimum measured wall thickness and reduce this by 10% to give minimum allowable wall thickness for calculating allowable heat input. (It was assumed that ultrasound is liable to overestimate by approximately 10% the wall thickness in an area of internal corrosion.)

  3. Fit sleeve halves to pipe and weld both longitudinal seams using backing strips to avoid penetrating into the parent pipe.

  4. Weld the circumferential fillet welds in three passes.

    The first pass is the root pass made with the stringer bead technique without weaving or backstepping. The second pass is made with the same technique as the first and placed next to the first pass and on the carrier pipe.

    The third pass is made on top of the first two passes joining them to the sleeve. This pass can be made at a higher heat input than the first two passes and some weaving can be employed as long as the carrier pipe is not remelted.

  5. Inspect all welds visually for poor profile undercut, porosity, etc., and repair if necessary.

  6. Carry out magnetic particle inspection of all longitudinal and circumferential fillet welds.

    With this procedure, sleeving of all three-star events was successfully and safely carried out during 1990 and 1991.

WRAPPING SYSTEMS

The replacement pipe selected for the project was the factory-coated epoxy/extruded-polyethylene system. This pipe coating had proven successful in the first reinstatement exercise.

Joint make-good for these pipes was initially specified as consisting of a double butyl laminate innerwrap and a butyl/polyethylene outerwrap. This system is extensively used for both line wrapping and wrapping by the length in a factory or yard.

Doing joint make-good on site, however, yielded mixed results. Problems occurred with tension, overlap, and adhesion, primarily between the tape and the factory coating on the pipe (Fig. 5).

A "controlled environment" test was conducted in the yard at site, culminating with the head office quality-assurance manager dousing the joints thoroughly with a fire hose to check for water penetration.

Results were inconsistent, although no water penetration was observed.

In view of these results and the practical difficulties of applying the tape system on site, it was resolved that the joint make-good system would be changed to the polyethylene/bitumastic system.

The wrapping system selected for the sleeved sections of the pipe was again the polyethylene/bitumastic tape with a 55% overlap, effectively giving a double wrap.

Two road crossings were replaced as part of the new pipeline installation. These were installed with the "cut and fill" method, not something that can be done in many developed countries.

The coated carrier pipe was first threaded through the casing. The casing was double wrapped with polyethylene/bitumastic tape and overwrapped with adhesive polyethylene.

The ends of the casing were sealed with link seals and case seals, with a combined vent-drainage pipe on the lower end of the casing.

The entire assembly was then lifted and installed in the ditch.

From a practical point of view, the polyethylene/bitumastic tapes appear to be superior to the polyethylene/butyl laminate tapes for site make-good and short sections of wrapping in the trench.

Their performance is more uniform and reliable, and the tapes are more forgiving in the hands of semi-skilled operators. One of the major factors affecting use of polyethylene/butyl laminate tapes is their sensitivity to dust contamination.

CATHODIC PROTECTION

The different wrapping systems along the pipeline vary considerably with respect to absolute wrapping integrity and inherent properties. This gives rise to wide variation in current density requirements to achieve cathodic protection and also to the length of pipe that can be protected from a single transformer rectifier (TR).

A recent potential profile of the line (Fig. 6) (12059 bytes) clearly shows the location of the various TRs. The TR spacing varies from 10 km in the areas near the coast, which were once rice and sugar cane fields and still have the original wrapping system, to some 40 km on the rewrapped/replaced section in the low lying areas up to 120 km in from the coastline.

Once the pipeline leaves the coastal flats and enters more mountainous terrain, CP becomes more difficult due to terrain that varies from totally noncorrosive sandy soil to clay pockets in rocky sections.

For the new pipe, this presents no problem, but for the original wrapping with its relatively low specific resistance and high current density requirements, conventional limits for cathodic protection sometimes must be exceeded.

The usefulness of the overline polarized potential survey in these situations has been demonstrated, in that "Off" potentials were found to be within the - 1.2v wrt CU/CUSO4 limit.

The new sections of pipeline laid in the latest rehabilitation exercise were electrically isolated from the old pipeline at each end in order to facilitate commissioning of the CP system and prevent possible overprotection of the new sections.

It was found that the short section of 10 km did not need to be insulated, and a bonding link was provided whereby the new line could be disconnected should the adjacent TR on the old line malfunction for any length of time. In this instance, the temporary sacrificial anodes could be reconnected.

On the other section of 30 km, an existing TR was used to protect the section independently. The current demand on the line was so low that the shunt on the TR had to be replaced with a 1-ohm resistor to yield adequate resolution on the instruments.

The original TRs were installed when access was difficult and power supplies could not be guaranteed. The units were therefore made to function under feedback control giving constant pipeline potential. This provided backup from adjacent units should one unit malfunction.

Variations in system resistance due to seasonal fluctuations were also catered for. In practice, the compensation system only worked on the new sections of line; the rest were operating at the limit of their protective ranges.

These TRs were susceptible to power and lightning surges (Fig. 7), and long-term stability of the reference electrodes in the fluctuating seasonal conditions was a problem.

The units were therefore stripped and rebuilt as constant voltage supplies, using blind thyristor controllers. The surge-protection design was also upgraded to provide better physical layout of the components.

This revamp has been successful, with the only alteration in operation being seasonal adjustment of the TRs.

PRACTICAL PROBLEMS

Repairing a 10-in., Grade X-46 pipeline presents very few problems in developed countries; in developing countries it can be a very different matter.

An unfortunate political situation in the country in question caused some problems with respect to the security of both personnel and equipment. Working was limited to daylight hours and the pipeline was guarded 24 hr/day by regular troops.

Obtaining the licenses required to import the relevant equipment and materials for the sleeving operations was difficult but eventually overcome.

The national government required maximum use of local labor. The absolute minimum expatriate labor was therefore employed on tasks that could not be performed by locals, such as welding, inspection, and installing cathodic protection.

The majority of the pipe sleeving was undertaken in the dry season and in a drought.

Paradoxically, there was no shortage of water at 6 ft below ground level, that is, at the bottom of the trench where the pipe lay (Fig. 8). This water caused a number of operational problems.

The pipe had to be sleeved in situ and therefore the water had to be expelled from the trench. This required extensive pumping and use of rafts to form a platform on which personnel could work.

The availability of water in a land of drought also attracted an undesirable number of local fauna during the night, when sleeving operations were suspended. Chief among the visitors were rodents and frogs. These became trapped in the trench and attracted an astonishing variety of snakes, all aggressive.

Hence the first task every morning was to evict the snakes from the trench to allow work on the pipeline to commence.

Laying the new section of pipeline did not encounter the problem of reptiles because the line was fabricated on the right-of-way. It did, however, pose some problems of its own.

Digging the new trench was frequently hampered by areas of rock. The tools available to remove the rock were few. Explosives were in very short supply.

Various methods were employed to trench through the rock including the tried and trusted methods adopted by Hannibal in his crossing of the Alps. This consisted of building large fires around the rock outcrops and letting them burn for several days. Cold water was then pumped onto the hot rock. This technique was not entirely successful.

More successful was the use of homemade explosive. Weedkiller was mixed with diesel; the resulting mixture was poured into deep holes drilled in the rock. This mixture proved reluctant spontaneously to explode from conventional detonators.

The solution was to set it off with a small amount of the precious "proper" explosives available. This technique was spectacularly more successful.

Another problem encountered during the trenching phase of operations was that the local inhabitants had decided to live close to the project in order to be under the troops' protection. This was a completely illegal occupation of land owned by the operating company but in the circumstances understandable.

The right-of-way hence carved a path through inhabitants' encampments. Once compensated, however, residents endured the destruction of their properties with equanimity.

Problems also existed where the pipeline crossed a major railway line. For strategic reasons, the railway did not operate a regular timetable. Thus, it was difficult to arrange a time to trench across it that was acceptable to all parties.

The solution was to complete the operation as quickly as possible.

Men with red flags were spaced out along the line from the crossing to wave the flags when a train approached. Fortunately this operation was completed successfully before the effectiveness of the safety arrangements was put to the test.

HYDROSTATIC TESTING

Obtaining sufficient water for hydrotesting is rarely a problem in developed countries and the chemical composition of test water is only of concern if high chloride levels would cause corrosion in stainless steels.

In developing countries, on the other hand, the availability of sufficient dean water for hydrotesting can be a significant obstacle. Obtaining sufficient water to fill the pipeline for testing was overcome by damming a local stream for several weeks (Fig. 9).

This practice dictated the lengths of line that could be tested. The open end had to be relatively adjacent to the water supply.

There were concerns about the general water quality and that the water would hold large amounts of mud.

Samples of the water were sent for analysis which showed that it contained sufficiently high levels of coliform bacteria to endanger both site personnel and the pipe itself.

Contact with this type of water can lead to dysentery, typhoid, or salmonella.

This concentration of coliform organisms is also associated with high levels of sulfate-reducing bacteria (SRBs) which could cause internal corrosion of the pipeline. The mud in the local river water proposed for hydrostatic testing would settle down on the lower quadrant of the pipe, thereby increasing the extent of localized corrosion damage, particularly in the presence of SRBS.

In a western-style operation, it would be recommended that the source water should be filtered and treated with both an oxygen scavenger and an organic biocide when used for hydrotesting. This approach was impractical for this project.

Considerable care was taken in evacuating the hydrotest water from the pipeline after testing to avoid the corrosion risks stated above. Several passes were conducted with a variety of pigs to ensure that the line was thoroughly cleaned and dried.

THE FUTURE

Work is now under way further to upgrade the pipeline in order to take increased pressure from new pumping stations. Another intelligent pig survey was undertaken last year that used a high resolution magnetic-flux tool.

Analysis of the results has yet to be completed but areas requiring further attention due to corrosion will be identified. New pipeline sections with modern protective coatings will overcome most of the remaining weak points in the cathodic protection/wrapping system.

The feasibility of installing small CP booster stations at critical areas on the line is also under assessment.

ACKNOWLEDGMENTS

The authors acknowledge contributions made to this article by colleagues at John Brown E&C Ltd. and permission to publish by directors of John Brown.

Copyright 1995 Oil & Gas Journal. All Rights Reserved.