An automated aid for isolating defects in the heat exchanger tubing of nuclear steam generators

The inspection of heat exchanger tubing in nuclear power plants, for purposes of detecting, classifying and quantifying defects, is presently accomplished manually by visual analysis of eddy current waveform data. The procedure is time consuming, costly and subject to human error. The objective of this study was to demonstrate the feasibility of using software as an automated aid in the detection of anomalous eddy current waveforms. Due to the difficulty of detecting defects in the presence of the support plate signals, the study concentrated on the detection of defects in the support plate regions of the heat exchanger tubing. The key purpose of the automated aid is to eliminate normal support plate response (SPRs) from the inspector's visual consideration. Thus, the purpose of the algorithm is to detect normal SPRs and dismiss them from the inspector's visual inspection. Conversely, the algorithm flags any anomalous waveform, whether the anomaly represents a defect or a benign phenomenon such as probe wobble. The inspector is still the decision-maker with regard to whether the anomalous SPR represents a defect or a benign event. A database of simulated pits of varying depth, diameter, and volume was obtained by pulling an eddy current probe through a once-through steam generator (OTSG) mock-up. The discriminator was successful in eliminating over 80% of the normal SPRs from the analyst's consideration. A the same time, almost all of the defects with a volume greater than 30,000 mils cubed or a depth greater than 10 mils were labelled anomalous and flagged for analyst interpretation. The authors conclude that this research demonstrates the feasibility of automated aids for the isolation of areas in steam tubes which may contain defects. An instrument that implements such an aid would reduce the time required to perform the general defect detection survey by approximately 75%. J