Gamma-ray Emission Tomography: Modelling and evaluation of partial-defect testing capabilities

Assessment of gamma emission tomography (GET) for spent nuclear fuel verification is the task in IAEA MSP project JNT1955. In line with IAEA Safeguards R&D plan 2012-2023, the aim of this effort is to “develop more sensitive and less intrusive alternatives to existing NDA instruments to perform partial defect tests on spent fuel assemblies prior to transfer to difficult to access storage". The current viability study constitutes the first phase of three, with evaluation and decision points between each phase. Two verification objectives have been identified; (1) counting of fuel pins in tomographic images without any a priori knowledge of the fuel assembly under study, and (2) quantitative measurements of pin-by-pin properties, e.g. burnup, for the detection of anomalies and/or verification of operator-declared data. Previous measurements performed in Sweden and Finland have proven GET highly promising for detecting removed or substituted fuel pins (i.e. partial defects) in BWR and VVER-440 fuel assemblies even down to the individual fuel pin level. The current project adds to previous experiences by pursuing a quantitative assessment of the capabilities of GET for partial defect detection, across a broad range of potential IAEA applications, fuel types, and fuel parameters. A modelling and performance-evaluation framework has been developed to provide quantitative GET performance predictions, incorporating burn-up and cooling-time calculations, Monte Carlo radiation-transport and detector-response modelling, GET instrument definitions (existing and notional) and tomographic reconstruction algorithms, which use recorded gamma-ray intensities to produce cross-sectional images of the source distribution in the fuel assembly or conclusive pin-by-pin data. The framework also comprises image-processing algorithms and performance metrics that recognize the inherent trade-off between the probability of detecting missing pins and the false-alarm rate. Here, the modelling and analysis framework is described and preliminary results are presented.