Application of Bayesian Inversion Techniques on Multi Frequency Eddy Current Data for Sub Surface Crack Profiling In Steam Generator Tubing

Steam generator (SG) tubing in nuclear plant offers a barrier between the radioactive material and steam. Effective NDT of the SG tubing is mandatory to ensure the safe operation of nuclear power plants. Multi-frequency eddy current (EC) NDT is being exercised in the nuclear industry. Detection and subsequent crack profiling from the acquired EC signals is a challenging task once dealt with subsurface and OD cracks. Data fusion based EC signal inversion techniques are proposed to address the gap. Measurement data was acquired from the simulation of sub surface cracks in steam generator tubing at three different scan frequencies. The proposed spatial correlation data fusion technique has been applied on the acquired multi frequency EC data. Flaw profile prediction from single frequency and data fused results has also been performed by using recursive Bayesian estimation procedure based on Sequential Monte Carlo method. To investigate the efficacy of the proposed data fusion method; quantitative analysis has been conducted. The robustness of the proposed technique was also assessed injecting artificial noise in the EC measurements.