DIRECT ASSESSMENT-Conclusion: ECDA tunes Gasunie corrosion predictions

Oct. 9, 2006
External corrosion direct assessment (ECDA) can serve as a powerful tool in determining not only the current but also the future health of a given pipeline system.

External corrosion direct assessment (ECDA) can serve as a powerful tool in determining not only the current but also the future health of a given pipeline system. Gasunie has developed software (PIMSlider) to aggregate ECDA data and allow increasingly accurate projections of line-failure probability.

Part 1 of this article, presented last week, provided the overall integrity management (IM) context in which PIMSlider is being applied. This concluding part examines the direct examination and postassessment steps of the IM process in greater detail and provides an ECDA example from Gasunie’s own system.

Direct examination

The direct examination step evaluates all inspected ECDA regions sequentially, starting with the highest-risk ECDA region found in preassessment. The outcome of each determines whether additional excavations need to be carried out, after which pipeline integrity is evaluated again.

The priority list generated by the two aboveground surveys allows selection of bell-hole excavation sites.

Excavations generally take place at locations:

  • Where both surveys (coating and corrosion) have given an indication; usually to determine the size of corrosion defects and repair any critical defects.
  • Where only one of the surveys has given an indication; usually to check the survey characteristics and determine the size of any found corrosion defects.
  • Where no indications were given; these excavations are generally referred to as blind digs, and may be used to assess the confidence in the probability of detection of the survey technique(s).

Information found at the excavations updates the information used in the preassessment and indirect inspection steps. This updating includes updating survey characteristics, number of defects, defect depth, corrosion rate, time of initiation, defect length, and the critical defect depth. These updates provide the basis for calculating new values for the probability of failure both of the defect and per km of the ECDA region.

The bell-hole excavations must gather information on: location of coating and (active) corrosion defects, dimensions of corrosion defects (defect depth, defect length in axial direction), and clock position of the corrosion defects (not relevant to the current model). Information should be used from every excavation, including those where no or only small defects were found.

All ECDA regions must perform model calculations. The direct-examination step makes the following groups of calculations (updates) for each ECDA region:

  • Survey characteristics.
  • Number of defects (coating and corrosion) as a function of time.
  • Defect depth, corrosion rate, and time of initiation.
  • Defect length and critical defect depth.
  • Probability of failure (of a defect and per km of the ECDA region).

The updating process for the number of coating defects is represented by a normal distribution. The prior estimate uses information collected during the preassessment step (e.g., coating condition, age of the pipeline), but may also be deduced from previous surveys on other similar pipelines. New information from the indirect inspections uses Bayesian statistics to immediately update the prior distribution. The excavations allow an update of the survey characteristics, after which the number of coating defects can be updated again.

The probability of failure increases with the growth of corrosion defects and an increasing number of defects, but generally decreases as a result of the indirect inspections and direct examinations. The probability of failure, however, will increase again due to ongoing corrosion processes.

ECDA assumes that all found corrosion defects are repaired (or recoated). The repair of coating defects will not have a significant effect on ECDA calculations, but assumes for consistency that all coating defects found are repaired.

The updated probability of failure of a defect at the time of excavation depends on the updated value of distribution of the defect depth, distribution of the critical defect depth, and expected number of corrosion defects. The updated probability of failure/km at the time of excavations depends on the updated distribution of failure of a defect and the length of the ECDA region.

The probability of failure for a single defect will change with each excavation as the distribution of the corrosion rate and time of initiation also change. A confidence interval is calculated for the probability of failure. As the number of excavations increases, the probability of failure/km normally decreases, until the criterion for the probability of failure is met, showing that sufficient excavations for the specific ECDA region have been carried out.

As mentioned earlier, excavation of ECDA regions occurs sequentially, starting with the region with the highest initial risk.

Results of the calculations for an ECDA region can provide the starting point for the next ECDA region. The user can replace the initial values for corrosion rate, time of initiation, and defect density for that ECDA region with the values calculated for the ECDA region with the higher initial risk. Inspectors repeat this process until all ECDA regions have been covered.

Postassessment

According to the NACE, the objectives of the postassessment are to define reassessment intervals and assess the overall effectiveness of the ECDA process. For the developed structural reliability analysis (SRA) model, this step consists of:

  • Calculating the future probability of failure/km due to external corrosion for each ECDA region, based on indirect inspections and direct examinations.
  • Calculating the future probability of failure/km for each ECDA region for all other considered failure modes.
  • Depending on the calculated probability of failure/km, calculating the time interval until the next direct assessment is required.

When new aboveground surveys are carried out on an already inspected pipeline the pipeline becomes subject to a new direct assessment. The results of a previous direct assessment are available for use in the preassessment step of the next direct assessment.

Click here to enlarge image

Fig. 1 gives a summary of the most important routines that the operator will go through when using the DA module for PIMSlider.

Case study

Table 1 contains the details of a pipeline subjected to ECDA in 2005. Assessing all relevant input parameters related to pipeline geometry, material properties, defect dimensions, and incident frequencies, including their respective uncertainties, yields a first estimate of the current integrity of the pipeline.

Click here to enlarge image

The first section of Table 2 shows this estimate: a prior failure probability per km of pipeline in the year 2005 (52 years after commissioning) of PF = 0.18/km; unacceptably high according to ASME B31.8.

PIMSlider identifies ECDA regions by retrieving and graphically displaying data necessary for preassessment (Fig. 2).
Click here to enlarge image

Fig. 2 gives an example of how PIMSLider can be used to assist a pipeline operator in performing a preassessment, displaying the geographic position of the pipeline under consideration, allowing the user to zoom in on any segment of interest, and showing corresponding pipeline, environmental, incident, or operational data. If the required data are missing for a specific pipeline, the DA module can retrieve data from similar pipelines to estimate the prior condition of the pipeline. The DA module also identifies and calculates the ECDA regions, following user definition of the relevant parameters and criteria to be considered.

In the indirect inspection step, the DA module allows the user to store and analyze data from aboveground surveys, assess the severity of defects, and identify areas where corrosion may occur (Fig. 3).

Click here to enlarge image

The last section of Table 2 shows the postassessment effect of the direct-current voltage gradient (DCVG) survey on the expected number of coating defects. The DCVG survey found 842 coating defect indications. Bayesian updating of the number of coating defects results in an increase of the expected number of coating defects from 713 (52/km) to 957 (70/km), taking into account DCVG’s initially estimated probability of detection and probability of false indication.

Applying Bayesian statistics also updates the number of corrosion defects.

The results of the cathodic protection survey showed that both the on and off-potential of the pipeline meet the applicable protection criterion described in European standard EN 12954:2001. Calculation of the IR-free potential, however, revealed 35 coating defect indications without sufficient CP protection and therefore possibly corroding, increasing the expected number of corrosion defects from 6.2 to 58.9.

Click here to enlarge image

The direct examinations step performed 21 excavations, all at DCVG indications. Each excavation found actual coating defects, allowing the probability of false indication of DCVG to be decreased to 4% from an initial value of 10%.

The absence of excavations at sites without DCVG indications (e.g., blind digs) prevented any update of the probability of detection.

The combination of these adjustments updated the expected number of coating defects after excavations to 1,076 (e.g., 79/km).

Fig. 4 shows the effect of the DCVG survey and the direct examinations on the probability density function of the number of coating defects.

Click here to enlarge image

The 21 excavations included eight excavations at sites where CIPS data indicated insufficient protection by the CP system. Actual corrosion anomalies were found at four of these sites. This information updated the probability of false indication of CIPS to 38% from 20%.

The other 13 excavations, where no corrosion was suspected based on the CIPS data, resulted in two additional corrosion defects, leading to a decrease in the probability of detection of CIPS to 27% from its initially estimated value of 60%. The updated performance indicators (probability of detection and probability of false indication) of both indirect inspection tools led the expected number of corrosion defects to increase to 181 from a prior value of six.

Accurate measurements of corrosion defect depths and lengths during the excavations allow an updating of their respective distributions. Updating the defect depth distribution allows calculation of the corresponding corrosion growth rate (Fig. 5).

Click here to enlarge image

Table 2 shows that the corrosion growth rate decreases significantly to 0.015 mm/year, as compared to its prior value of 0.054 mm/year, resulting in a much lower probability of failure as compared to the preassessment.

ECDA, in combination with SRA, effectively demonstrated the integrity of the pipeline. Even though the expected number of corrosion defects is much higher than estimated initially, the corrosion anomalies found were all minor in nature, resulting in a significantly lower average defect depth than initially estimated.

Click here to enlarge image

Fig. 6 shows the overall effect of the ECDA process on the probability of failure. The adjusted probability density function of the number of corrosion defects, together with the updated defect depth distribution and the updated critical defect depth distribution, result in a decrease of the probability of failure to PF = 4.0 × 10-9/km.