Verification of split spectrum technique for ultrasonic inspection of welded structures in nuclear reactors

Ultrasonic nondestructive inspection of materials is often limited by the presence of backscattered echoes from the material structure. A digital signal processing technique for removal of this material noise, referred to as split spectrum processing (SSP), has been developed and verified using simple laboratory experiments during the last decade. However, application of the split spectrum processing algorithm to industrial conditions has been rarely reported. In the paper the results of the practical evaluation of the SSP technique are presented. A number of different ultrasonic transducers were used for acquiring echoes from artificial flaws as well as natural cracks. The flaws were located in test blocks employed by the Swedish Nuclear Power Companies as reference during ultrasonic inspection of nuclear reactor vessels. The acquired ultrasonic A-scan signals were processed off-line using specially developed algorithms on a personal computer (PC). The experiments show evidence that properly tuned SSP algorithms result in a considerable improvement of the signal to material noise ratio. The enhancements were similar irrespective of the features of the transducer used or the nature of the inspected flaw. The problems related to the development of self-tuning SSP algorithms for on-line processing of B-scans are discussed. (author)