124
Advances in API 1163 Level 3 ILI Validations
1T. D. Williamson, Salt Lake City, USA. 2Integral Engineering, Edmonton, Canada
The second edition of API 1163, In-line Inspection Systems Qualification, was published in 2013 and introduced the idea of Level 3 in-line inspection (ILI) validation, where field measurements are compared to ILI measurements to arrive at an as-run assessment of ILI performance. However, it provided little concrete guidance on how to conduct such a validation. The third edition of API 1163, published in 2021, made some headway on this by presenting a brief overview of two approaches to Level 3 validation using either statistical tolerance intervals or Bayesian regression.
Much has changed in the five years since the third edition was published! Operator experience with both Level 2 and Level 3 validations sparked two PRCI projects aimed at providing additional insight into the behaviors of the techniques and guidance on their application. We present an overview of recent advances in Level 3 ILI validation techniques specifically designed to address performance questions that are difficult to answer by Level 2 techniques, including:
- What is the “best estimate” of how the ILI system performed in this pipeline run?
- What is a conservative estimate of performance that bounds all the field uncertainty and the sampling uncertainty?
- Can we remove field measurement uncertainty from the ILI performance estimate?
- Can we apply a linear calibration fit as part of the estimate?
- Why do the Level 2 and Level 3 techniques provide different guidance, and what are the implications?
Both frequentist and Bayesian approaches are discussed, including brief application examples.
Available after the event



