Sensors and Instrumentation and Nondestructive Evaluation
Safety Related Applications
DOE Office of Nuclear Energy, Science, and Technology
The objective of this Nuclear Energy Plant Optimization Project is to improve the analysis of eddy current data for axial and circumferential stress corrosion cracks in steam generator tubes. The high-resolution eddy current probe that was selected for this study is the RG3/4, which is a surface-riding transmit/receive (T/R) probe developed by RDTech. Inherently low lift-off noise and an absolute type response are the primary characteristics that make this directional probe a viable candidate for flaw sizing. Inspection data were acquired with RG3/4 using the steam generator mock-up at Argonne.
By combining the effort to increase signal-to-noise ratios for the X-Probe array (the RG3/4 is a rotating element of the X-Probe array), along with enhanced flaw imaging and sizing algorithms from a program sponsored by the Nuclear Regulatory Commission (NRC), a powerful capability is being developed for characterizing and sizing cracks in the tube sheets of steam generators. The RG3/4 data will be processed by computer algorithms developed at Argonne to produce NDE profiles.
Evaluation of the eddy current probe will continue using field-degraded tubes removed from the steam generator of the McGuire Reactor under NRC sponsorship and data provided by the Electric Power Research Institute (EPRI) from their qualification program for the RG3/4. Eddy current data from radioactive field samples are being collected for this evaluation using the NDE glovebox at Argonne.
Conventional denoizing techniques such as wavelet transformations require some a priori knowledge about the nature of the degradation. This information is not always available from eddy current data, which can limit the use of such techniques for automated data analysis. Blind source separation (BSS) techniques, on the other hand, seem to be ideal for noise with strong spatial variability, which often appears in conjunction with eddy current signals. Such signal processing methods allow extraction of additional information from already available eddy current data. Conventional independent component analysis techniques assume a linear mixing of signal and noise. However, this may not always be the case for eddy-current inspection results in the presence of deposits and support structures. We are investigating novel nonlinear BSS methods, which provide a more powerful and general methodology in dealing with eddy current sources of noise.
Last Modified: Mon, April 19, 2010 3:14 PM