Maximizing flow capacity

Over the last few years, Brazil has noticeably increased its oil production and theoretically achieved self-sufficiency. The major new production fields discovered contain reservoirs of heavy oil, whereas the country’s refining and transporting infrastructure had mostly been planned and constructed considering a scenario of imported light oil. Thus, it has become a strategic matter to revamp the national crude oil pipeline network, in order to accommodate the conveyed fluid viscosity boost.

However, as a result of their previous service lives, some of those systems have accumulated considerable corrosion damage. In order to guarantee future serviceability of these particular assets, PETROBRAS Transporte has developed a special methodology designed to increase the crude oil delivery system reliability and, simultaneously, minimize costs related to the rehabilitation of pipelines submitted to substantial upgrading, especially those severely corroded. A sine qua non directive of the current project was not to compromise traditional safety margins, dictated by engineering best practices.

Theoretical background

Corrosion is the natural tendency of a metal to revert back to its more stable state as an ore. By reason of it being submitted to internal and/or external environmental aggression, buried steel pipelines eventually exhibit metal-loss damage. Indeed, corrosion has been a major threat to pipeline longevity and reliability, and most of their operating and maintenance costs are related to the control and monitoring of the process evolution. The main concern is its effect on pipeline structural integrity and, therefore, extending the life of aging pipelines and/or design parameter changes ought to take proper account of metal-loss areas assessment.

To determine if particular corrosion damage should or should not be submitted to a repair, a direct comparison between service and allowable pressures must be performed. This could be expressed by the following ratio, commonly referred to as the “Estimated Repair Factor” or ERF.

ERF = MOP/P_allowable     (1)

Throughout the last half-century, much effort has been devoted to predicting failure in corroded pipeline areas, resulting in, firstly, the development of a range of criteria to determine the remaining strength in metal-loss areas and secondly, the evolution of ILI technology aiming at damaged site measurement and mapping. However, the above achievements should only have an effect on the denominator of Equation (1) right hand term. As a matter of fact, traditional approaches consider MOP as an unchanging value along the whole length of the analyzed pipeline segment but, depending on the pipeline’s ROW topography and relief system configuration/set-ups, the ability to accurately determine local pressure worst-case scenarios could play a major role in the definition of rehabilitation points. To avoid over-conservatism associated with traditional fitness-for-purpose procedures, detailed MOP gradients could be constructed by means of hydraulic simulations. The results could then be used to perform defect assessments, considering the worst-case scenario for local pressure at each damaged site.

The assessment methodology

Hydraulic simulations were carried out for the most relevant transporting scenarios (including diverse batching products; start-up and shut-down of pumping station/unit; different alignments and configurations, and valve blockage scenarios). Pipeline STONER Simulator 9.3 was used to perform the simulations, it being appropriate to mention that resultant pressures associated with transient state took into consideration surge allowances.

As severely corroded pipelines submitted to the scrutiny of the latest high resolution ILI tools could result in vast metal-loss anomaly populations (as high as tens or sometimes hundreds of thousands) the proper way to assess them is by computational means. Accordingly, the company’s proprietary defect assessment package – PLANPIG 2.0 – was used. The former framework comprised independent deterministic and probabilistic approaches, regarding current and future integrity analysis, respectively.

Remedial actions are defined considering tool measurement error and corrosion rate as random variables; assuming Gaussian distributions to model their behavior. Re-inspection intervals are ascertained by means of a risk-based inspection (RBI) methodology. The software also defines acceptable failure probabilities through the solution of an analytical equation considering defect growth, ranking possible re-habilitation repair scenarios based upon their cost-effectiveness. Additionally, in order to provide a broad-spectrum cover regarding the peculiarity of service conditions and defect geometries, a range of criteria has been implemented:

• Defects longitudinally oriented under internal pressure load (ASME B31G, RPA)
• Defects circumferentially oriented under internal pressure load (RPA-PLLC)
• General defect under thermal and pressure combined loads (Kastner).

On the subject of metal-loss evolution over time, current work has considered distinct tri-dimensional growth rates, automatically determined, based upon categorized defect geometry parametric relationships. To define mean corrosion rate input, a variety of methods were employed, such as: corrosion monitoring techniques, field-collected data (electrical resistance probes and/or weight loss coupons), successive correlations conducted in the same spot, statistical studies of single/multiple pig-run data, and corrosion prediction software runs. Conversely, the corrosion rate variance is assumed to be a constant value, which was defined by studying typical probabilistic distributions on the company pipeline defect database.

In the case of huge operational changes, it is not uncommon that hydrotest is required to demonstrate the ability of the system to withstand the new requirements and, depending on segmentation strategy; this could also have a considerable impact on the repair numbers for pipeline rehabilitation. To eliminate over-conservatism at this stage, special considerations were also given to assess defect population under these particular conditions: the adopted approach was the mathematical admission of plastic yielding on the corroded areas. Despite that, real phenomenon is not a common occurrence in practical applications – the reason is the usual conservatism imbibed on standard criteria. It is worth noting that defect growth in the time interval between pig inspection and the hydrotest execution was considered.

Case study 1

This is a trunk line with constant diameter and thickness (22 in. x 6.35 mm) connecting a refinery to its coastal facilities located 100 km away. Most of the refinery crude oil supply flows through this pipeline, constructed using two grades of line pipe (API 5L X46 and X56). It was commissioned in 1969, but from 1999 to 2007, the pipeline endured critical operational conditions: with recurrent production water pumped together with high BSW content products and intermittent movement. Long standoff periods were a regular occurrence. The internal corrosion monitoring device data were available (electrical resistance and weight loss coupons) as was past data for over line surveys (DCVG, PCM and CIPS). At the beginning of 2007, the conclusion of a refinery revamp project caused the pipeline a demand adjustment to higher viscosity fluid transportation, together with a flow increase.

When a high-resolution magnetic flux leakage pig-tool was used to inspect the pipeline late in 2005, the results reported 203,652 anomalies, the majority located on the internal pipeline surface, between the 5 and 7 o’clock positions. This fact had been considered consistent evidence for the hypothesis that the severe internal corrosion process taking place at the pipeline bottom was due to free water presence. Figure 1 shows a pipe section where the severity of the preferential corrosion at the pipe bottom is very clear.

In order to achieve the new flow demands, the main station’s pumps were replaced and a new intermediate station constructed. Hydraulic simulations have been performed considering two basic configurations:

• Main pumping station operating alone: representing old operational conditions (Equation 2)
• Main pumping station together with the intermediate one: representing new required conditions (up and downstream intermediate pumping stations are explicitly stated by Equations 3a and 3b, respectively).

Liquid column energy lines have being graphically plotted on the chart, Figure 2, which illustrates the large differences between the old and the new operational profiles.

 (2)

(3.a)

 (3.b)

Where

P_i is the pressure at each assessed point, in [kgf/cm²]

H is the pipeline elevation, [m]

L is the pipeline length at each assessed point [km].

A new hydrotest was conducted in a single section. In order to predict the defect geometries at the test execution, a corrosion growth of 18 months (time interval between the smart-pig inspection and the test execution) was considered. It must be emphasized that despite the hydrotest rehabilitation threshold being superior to the operating forecast ultimate-pressure envelope most of the time, only the conditions represented by the latter were used to define the repair scopes for the entire rehabilitation time interval of 5 years. Results from this case, as for the others, are summarized in Table 1.

Case study 2

With a 16-in. constant diameter and thickness of 7.92 mm, this line connects a refinery with its coastal facilities. It was constructed from API 5L Gr. B line pipe and commissioned in 1968. Since then, it has been used to transport naphtha and crude (also with high BSW content products). Production water transportation was a frequent occurrence. Operating was conditioned by the tankers’ arrivals, which meant that the pipeline faced standoff periods as long as 15 days. Internal corrosion monitoring device data were not available, but data from past over-the-line surveys were (DCVG, PCM and CIPS). Old operating data suggest souring in past movements.

When a high-resolution magnetic flux leakage smart-pig was used to inspect the pipeline in late 2006, the results reported 789,117 anomalies. The vast majority of those points were internal metal loss principally located between the 5 and 7 o’clock positions. Again, the water decantation, during standoff periods, was appointed as the major cause of the internal corrosion.

Given the deteriorated condition of this line; it was defined that it was not suitable for crude transportation anymore. As the new diesel refinery production level exceeds the local demand, and it was necessary to export it by sea, it was decided to use this line. To achieve the new demands a new pumping station was erected at the refinery, in order to provide the product flow towards the coast. Constructed MOP gradient for this new condition is represented by the Equation below, while Figure 3 presents the topography and the pertinent resultant heads. It is worth noting that previous hydrotesting (performed only four years ago) considered the scope of the new operational requirements:

    (4)

The forecast repair scenarios were constructed based on the following hypothesis: crude oil movement continuity limited to a one year period after which, the four subsequent operational years (considering a rehabilitation project scope of 5 years) were only for diesel flow.

Case study 3

This line has a 30-in. diameter with a thickness range of 10.3-14.6 mm, and connects a refinery to a distribution facility some 152 kilometers away. It was constructed using two grades of steel linepipe (API 5L X46 and X52) and commissioned in 1976, since when it has been used to transport crude. Internal corrosion monitoring device data were available. In order to achieve the new required operational conditions, two new pumping stations were erected and a pipeline section of five kilometers was changed for pipes of higher-pressure design.

ILI data available resulted from a high-resolution ultrasonic smart-pig inspection carried out in late 2004. The results reported 720 anomalies, of which 131 are external metal loss while 199 are internal. Ultimate-pressure envelopes were established considering the three pumping stations to be operating simultaneously. MAOP gradients are described by Equations 5a, 5b (respectively up and downstream of the first intermediate pumping station) and 5c (downstream of the second intermediate pumping station). Figure 4 presents the topography and pertinent heads of this pipeline:

 (5a)

 (5b)

 (5c)

Conclusions

Many aging liquid pipeline systems face demands to increase viscosity and/or flow capacity requirements. A methodology to assure old pipelines’ upgrading in serviceability was developed, comprising the following steps:

1. Worst-case scenarios test-driven by hydraulic simulations
2. Construction of MOP gradients
3. Defect population assessment considering appropriate failure model and MOP gradient
4. Hydrostatic test segmentation strategy definition
5. Re-assess defect population considering the hydrotest pressures.

The model’s essential input parameters were: ILI data, pressure gradients, and depth corrosion growth estimates. Despite major concerns being usually focused on the latter, it has been found that the former could also play a very important role, particularly for short-term forecasting. Case studies on pipelines provide ample illustration of overall methodology applicability by comparing: former and current MOP gradients, together with the hydrotest pressure profiles and segmentation logistics, and even presenting a comparison between resulting rehabilitation scenarios.

Acknowledgments

The authors would like to thanks PETROBRAS Transporte S.A. for permission to publish this paper, and their colleagues S.M.F. Augusto, R. D. de Souza and O.C. Gonçalves. Based on a paper presented at the ASME’s 7th International Pipeline Conference, held in Calgary, Alberta, Canada, September 29-October 3, 2008.