Skip to content
Validating ILI Performance with API 1163 and PRCI Research in CIM

Those tasked with a job function related to in-line inspection - from working for an in-line inspection (ILI) service provider to analyzing the ILI results to evaluating ILI-identified pipeline defects “in the field” - may be intimately familiar with the guidance document API 1163 In-line Inspection Systems Qualification published by the American Petroleum Institute. You may also be familiar with the challenges of interpreting AND implementing said document! 

To that end, the Pipeline Research Council International (PRCI) published guidelines to provide clarity on how to correctly implement API 1163 and even created a handy spreadsheet tool to assist with validating ILI performance. Much of the guidance document provides clarity on how to perform and calculate a Level 1, 2 or 3 analysis and when each level is applicable.  

As it’s OneBridge’s mission to assist pipeline operators in preventing pipeline failures by utilizing their assessment data better, incorporating the PRCI guidelines and calculations contained within the spreadsheet was a no-brainer. Indeed, the OneBridge team worked with a user group to discuss their needs for conducting an API 1163 validation within CIM and the associated workflow.  

In CIM’s upcoming release, users will be able to perform a Level 2 and Level 3 validation per API 1163 and PRCI’s project IM-1-06 and associated report entitled “API 1163 Performance Validation Guidelines” published on March 17, 2023 and re-released on May 08, 2023. In a future update, users will be able to perform a Level 1 “similar pipelines” analysis i.e. comparing ILI to ILI data on different but similar pipelines.  

Introduction to API 1163 

If you’ve ever heard the term probability of detection (POD) or seen +/- 10%, 80% in an ILI report, you can thank API 1163. Indeed, this (90 page!) guidance document not only describes how pipeline operators can confirm the quality of ILI systems but maybe more importantly, delineates how ILI service providers should qualify and communicate the performance of their technologies, which is especially important for new technologies. Let’s discuss in detail some of the more interesting aspects of this important guidance document.  

Probability of Detection (POD), Probability of Identification (POI) and Sizing Accuracy 

In Section 6 Qualification of Performance Specifications, it’s explained that the performance specification of an ILI system should clearly state the type of anomalies, components and characteristics that are to be detected, identified, and sized by the ILI system in industry-standard terms. As you may have guessed, the ILI system will have different detection capabilities for different feature types. Figure 1 graphically shows how metal loss indications can be classified.  

 Probability of detection (POD): Qualifies the probability that an anomaly will be detected. A detection threshold and POD must be stated for each anomaly type.  

  • Example: For an extended metal loss feature (qualifier/type – see Figure 1 below), the Depth Detection Threshold = 10%t and POD = 90%. (t = wall thickness) 
  • What does this mean? The ILI system will detect 90% of all extended metal loss anomalies that have a depth of 10% or greater.  
  • Did you know? PODs are not just for metal loss. Depending on the technology employed, an ILI system could have published POD specifications for cracks, deformations e.g. dents, pipe ovalities, wrinkles, etc., metallurgical features, girth welds, hard spots, and more!  

24-05-23 091337

Figure 1: Dimensional classes for metal loss indications, Figure 3 in API 1163 

Probability of identification (POI): the probability that an anomaly, component or characteristic is correctly identified, stated with a detection threshold and POI.  

  • Example: For an isolated crack found in the body of the pipe (qualifier/type), the Depth Detection Threshold = 1 mm and POI = 90% 
  • What does this mean? The ILI system will correctly identify (as a crack or crack-like defect) 90% of the isolated cracks in the pipe body that have a depth of 1 mm or more.   

05-23 091507

Figure 2: Characterizing Cracking Probabilities of Identification, Table 3 from API 1163 

Sizing Accuracy: characterizes how close the reported dimensions will agree with the true dimensions. This is described with a stated tolerance (e.g. +/- 10%t) and a certainty (e.g. 80% of the time). A sizing accuracy is typically reported for depth, length and width of an anomaly. 

  • Did you know? Pipeline operators tend to focus on sizing accuracy the most when validating tool performance via API 1163. 


API 1163 - 3 Levels of Validation 

The higher the level, the more rigorous the validation. 

  • Level 1: the performance specifications from the ILI service provider are utilized but neither proven nor disputed. A Level 1 “applies only to pipelines with anomaly populations that represent low levels of risk.” Anomaly evaluation/validation measurements are not required. 
  • Level 2: the performance specifications are tested by calculating a binomial confidence interval with 3 outcomes (see PRCI guidance document for more info.)  Validation measurements required 
  • Level 3: as-run performance is calculated. Anomaly evaluation data needed. Validation measurements required. 


Fun* Facts 

API 1163 

  • API 1163, Second Edition is now incorporated in the Code of Federal Regulations for both gas and hazardous liquid pipelines. Per §192.493 and §195.591 - an operator must comply with the requirements and recommendations of API 1163 when conducting an in-line inspection. 
  • API 1163 discusses the performance validation of the ILI “system” which is the ILI tool and the associated hardware, software, procedures, and personnel required for performing and interpreting the results of the ILI. So, you’re not just validating the technology but everything else associated with that tool run! 
  • There are multiple variables that can limit detection thresholds, PODs, POIs and sizing accuracies including anomaly shape, anomaly orientation angle and proximity to other anomalies or components. For example, a crack near the long seam will be harder to detect than a crack in the pipe body.  
  • What’s the difference between anomaly, defect, imperfection, and feature?
    • Anomaly: any unexamined deviation from the norm in pipe material, coatings, or welds, which may or may not be an imperfection, defect or feature?
    • Defect: a physically examined anomaly with dimensions or characteristics that exceed acceptable limits.
    • Imperfection: an examined anomaly that does not exceed acceptable limits.
    • Feature: any physical object detection by an ILI system. Features may be anomalies, components, metallic objects, or some other item.  

PRCI Guidance Document 

  • Pipe failures have shown that burst pressure capacity is most sensitive to anomaly depth uncertainty for both metal loss and crack-like anomalies. (This is why pipeline operators tend to focus on depth when validating ILI performance.) – Section 3.3 
  • The field measurement error should be small relative to the ILI measurement error. – Section 4.2. 

*Fun is used loosely. 

 API 1163 and PRCI Research Implemented in CIM 

As mentioned previously, the OneBridge team has collaborated with a user group to implement both API 1163 and PRCI’s guidance document into CIM, so that validating your ILI data can be completed within minutes. Conducting an API 1163 analysis is accessed via the Integrity Compliance process within CIM, similar to how a user would analyze ILI data to identify repair conditions.  The “Data” tab of the PRCI spreadsheet tool, seen in Figure 3 below, was recreated within CIM and all data fields mapped accordingly. The outputs found on the “Calculations” tab are then calculated to create the applicable unity plots with an example graph seen in Figure 4 below.  

 S2024-05-23 091636

Figure 3: Screenshot of the row headings contained on the “Data” tab of the PRCI spreadsheet tool. 

API fig 4

Figure 4: Example of an output generated from an API 1163 Level 2 analysis completed in CIM. 

API fig 5

Figure 5: Example of an output generated from an API 1163 Level 3 analysis completed in CIM. 

Stay tuned for more API 1163 analysis updates within CIM - coming soon! 

Comments or questions? We’d love to hear from you!