A critical component of any in-line inspection (ILI) campaign is validating the collected data using direct in-field measurements. This validation process is intended to confirm the dimensions and classifications of the reported ILI features, allowing for increased confidence in integrity decision making. In the United States, the process is a legal requirement under the Pipeline and Hazardous Materials Safety Administration (PHMSA); it is also seen as an industry best practice in many other countries around the world. However, the quality and reliability of these validations rely heavily on the competence of individual in-field personnel.
The published performance specification of a given ILI system provides a statistical basis for its detection, classification and sizing performance. These specifications are created and refined through extensive testing of representative anomaly samples in small-scale (laboratory trials), full-scale (pull-tests) and real-pipeline environments. In-field techniques do not typically have the same level of rigor applied to understanding system performance. User competence can have a significant influence on the tolerance of in-field inspections. Because it can vary significantly from one operator to another, in-field tolerances are generally ignored in the industry. This includes nondestructive testing (NDT) techniques and destructive methods like in-field grinding and buffing.
Features with a high criticality from an integrity perspective that are identified by ILI are usually selected first for in-field verification. This is to ensure that any critical repairs are undertaken in a timely manner to ensure the continued safe operation of the asset. A conservative repair threshold can be used to ensure that a defect does not need to be revisited, as the planning and excavation of features can mean significant costs in terms of both time and resources. In this case, the inspection tolerance is of lesser importance, as the inherent conservatism of the repair removes the need for a strict in-field tolerance.
However, for ILI validation purposes, minimizing in-field tolerance is of critical importance. Only a small number of in-field features are used to demonstrate the acceptability of, potentially, thousands of features. In order to validate ILI performance, a validation system with high accuracy and a known level of tolerance is needed to provide confidence in the ILI specification performance – and not the much larger combined tolerance of the 2 techniques as described in API 1163.
Understanding in-field tolerances has a significant impact on the validation campaign and future integrity of the pipeline.
UNDERSTANDING IN-FIELD TOLERANCES FOR VALIDATION
Unity plots are a key visual tool used in determining ILI tool performance. But are they being used correctly if the in-field tolerance is ignored? Only by understanding the importance of the combined tolerance in relation to the unity plots does the real importance of the in-field tolerance become apparent.
For external metal loss, the tolerance of a magnetic flux leakage (MFL) system can be readily improved upon, by an order of magnitude, by using high-precision in-field techniques such as micrometers or laser scanners (with a typical tolerance of approximately ±0.2 mm). Providing high confidence in the ILI measurement because the combined tolerance increase is small, this is typically seen in a unity plot where there is little or no increase in the combined tolerance. Any points that fall outside of this combined tolerance range are said to be outside of specification.
For ultrasonic crack detection (UTCD) or electromagnetic acoustic transducer (EMAT) ILI systems, the technologies used to validate them have a similar, if not potentially lower, level of accuracy and tolerance. They are very heavily influenced by the in-field technology selected, the user experience, the feature morphology and the process implemented. Thus, the influence on the combined tolerance is greater.
BS 7910 recommends a tolerance of ±3 mm for a conventional shear wave inspection of a crack where the individual inspector tolerance is unknown. For an immediate integrity assessment using the in-field measurements, this is sufficient, although potentially inefficient when it comes to implementing repairs. Contrary to logical reasoning, applying a large in-field tolerance is nonconservative for validating ILI performance because the confidence in the accuracy of the ILI sizing is decreased. As the combined tolerances increase, so does the potential to accept features that fall significantly away from the ILI specification. This is shown in the plot below, where the window of acceptability increases significantly as the field verification tolerance increases. Crack sizing variability between crack sizing systems can vary from 0.5 mm to 4 mm, as referenced in a recent PPM paper titled “Tolerance of ILI Validation Inspections, Why Is It Important, and How to Reduce It” (Oldfield et al., 2023).
The plot above shows that when ignoring in-field tolerance, the ILI appears to be out of specification with more than 20% of the features sitting outside of the specified ILI tolerance. Including a ±0.7 mm in-field tolerance contribution means the features are within 80% confidence of the combined tolerance bounding with high confidence in the ILI specification results. However, with a ±3.0 mm in-field tolerance, all points are within the acceptance criteria of API 1163, but there is low confidence in the sizing tolerance of the ILI specification, which is generally applied to all integrity calculations.
It is for this reason that understanding in-field tolerance is so critical: it has the potential to increase uncertainty over the ILI tool performance when in-field tolerances are high. Equally, it also has the potential to give much higher confidence in the results when the in-field tolerances can be reduced.
Specifying an in-field target tolerance that has a small to negligible effect on the combined tolerance should give confidence that it is mostly the ILI tolerance that has been met rather than a larger combined tolerance. The challenge, then, is to ensure that the technician can achieve the target tolerance. The best way to accomplish this is through a blind trial with representative defects on pipe of similar properties, including diameter and wall thickness and material, to ensure that the procedure and technology used are sufficient and, more importantly, that the technicians are competent enough to follow the procedure and meet the requirements of the target tolerance.
API 1163 defines the combined tolerance approach (shown in the equation below). From this, the desired in-field target tolerance can be back-calculated (as shown in the plot below for a fixed ILI tolerance of 1 mm). Here, an in-field tolerance of ±0.7 mm gives a combined tolerance of 1.2 mm. This is both a small influence on the combined tolerance and a realistic target for a skilled and experienced technician to achieve.
Understanding and minimizing the in-field tolerance is a critical component in validating ILI performance. With large in-field tolerances, uncertainty is cast over the ability of the ILI tool to meet its performance specification. While small tolerances offer increased confidence in the ILI performance in crack sizing validations, there is a tradeoff between what is realistically achievable in-field and minimizing the contribution to the combined tolerance. But understanding the tolerance of the individual in-field inspectors is a critical component of this task.
Guidance on the blind trial process and ensuring that the in-field technicians used to validate ILI performance are competent can be found in the ROSEN “Field Verification Requirements” document, which is issued with every master service agreement. Further guidance can be obtained by talking to a ROSEN subject matter expert about how to improve in-field data quality and ultimately pipeline integrity decision making.
Oldfield, T., Fowler, S. and Torres, D. (2023), “Tolerance of ILI Validation Inspections, Why Is It Important, and How to Reduce It,” Pipeline Pigging and Integrity Management Conference (Preprint).