A Comparison Study on Processing ILI Data With Different Filtering Methods

Abstract

In-line inspection (ILI) data processing is a crucial process on assessing the remaining capacity of corroded pipelines. It is used to generate the RBP (river bottom profile) or M AV (moving average) file which could be further used in the capacity calculations. There are large uncertainties regarding the qualities of ILI data itself, which are dependent on the accuracies of the ILI tool, the reporting method and the condition of the inspected pipelines et. al. It is thus required that the post-process of the ILI data should not add more uncertainties to the capacity check. Challenges exist on generating reliable RBP/MAV from the ILI data on general base. This paper originates from work during the in-house software development for corroded pipeline tool. Three different algorithms on filtering the ILI data are examined. They are the method recommended by DNV-RP-F101 (with modification), the median filter algorithm and the gaussian filter algorithm. The latter two algorithms are from standard methods for image denoising. A comparison study has been performed with several actual ILI data sets. Finally, conclusions and suggestions have been made, which may provide useful hints for industry application.