Influence of defect characteristics on inspection reproducibility of automated testing machines

Speaker:
Würschig, Thomas; GE Sensing & Inspection Technologies GmbH; Germany

Authors:
Würschig, T.; GE Sensing & Inspection Technologies GmbH; Germany
Breidenbach, C.; GE Sensing & Inspection Technologies GmbH; Germany
Hömske, B.; GE Sensing & Inspection Technologies GmbH; Germany
Pfortje, R.; GE Sensing & Inspection Technologies GmbH; Germany
Falter, S.; GE Sensing & Inspection Technologies GmbH; Germany

ID: ECNDT-0144-2018
Download: PDF
Session: NDT Reliablity 1
Room: J2
Date: 2018-06-14
Time: 09:00 - 09:20

The reproducibility of the inspection is one of the key figures that determine the performance of an automated testing machine. The measurement process applied during the commissioning of these machines is simple: according to valid standards and agreement between the parties, a number of inspection runs has to be carried out on the reference pipe, that has been used to adjust the machine. The measurement quantities per reference defect collected during these runs are compared to each other and show up with a characteristic variation This variation, on the one hand, it is governed by statistic variations giving rise to mostly Gaussian distribution of these variations around the average value. On the other hand, variations occur, that show up with non-gaussian probability influencing the reproducibility in a totally different, less homogeneous fashion. One of the major sources of non-gaussian variations is the quality of the reference defect manufacturing process. In this paper, we will show up with an in-depth investigation where we measure the reflection characteristics of the reference defects and compare those to the obtained reproducibility values for the same reference defects. We use these results to develop a correlation process that allows to estimate the influence of the measure reference-defect and define a formula that allows to predict an upper limit of the maximally obtainable reproducibility with a given reference defect.