The performance of inline inspection (ILI) tools has dramatically improved over the last two decades, in particular, with regard to detection capability and sizing accuracy of the respective inspection methods. Inline inspection results can hence directly be used as input parameters for defect assessment.
Assessment reliability however strongly depends on the accuracy of the input data. Performance characteristics of an ILI tool have to be defined in a quantitative manner that is usually stated in a defect specification.
Any meaningful specification requires the use of appropriate statistical measures. This is more and more recognized by the industry and has led to the creation of standardized guidelines. However, these guidelines seem not to be applied in a consistent manner yet.
The paper gives an overview of basic statistical concepts used in the field of ILI and explains the main ideas (probability of detection, confidence levels, sizing tolerances etc.) in their specific context, while demonstrating relevant examples. The paper/presentation also discusses statistical implications of the use of a combination of different inspection technologies.