Traditionally, nuclear forensics is described as a comparative science, where unknown materials are compared to sets of information in databases. When applied to spent nuclear fuel, the reliability of information in a database becomes embedded with uncertainty and inconsistencies arise among entry contents. In industry, spent fuel data is commonly reported in a very conservative manner for criticality and radiation protection, making poor usage for nuclear forensics purposes. In addition to conservativeness in reported information, a given reactor has great variance in its spent fuel's isotopic composition. Particularly with research reactors, non-symmetric designs, experimentation and operation histories produce a broad range of isotopic compositions in spent fuel. From normal usage, a reactor will produce spent fuel with a wide range of compositions that will be difficult to establish uniqueness in a database. Utilization of a spent fuel inverse analysis system enables the reconstruction of additional reactor data for usage in database comparisons that has significantly more credibility. If this system were to be applied on an unknown spent fuel sample, a representative subsample must be acquired and radiochemical analyses performed. The primary analyses consist of inductively coupled mass spectrometry (ICP-MS) and gamma spectrometry using a high purity germanium (HPGe) detector. The reactor information is then iteratively reconstructed using an inverse analysis system. This is composed of several reactor depletion codes that are used as a forward model in a constrained numerical optimization algorithm. The system reconstructs parameters involving the initial reactor fuel compositions and operation parameters. A minimization function is established by the Euclidean norm of spent fuel forensic signatures, comparing the iterative solution to the measurements from the spent fuel sample. As in any numerical inverse analysis, solution uniqueness and convergence are not guaranteed and are strongly dependent upon a reliable initial guess. The information recovered in this analysis, when utilized in a database search method, is much more reliable, widely available, and possesses lower uncertainty than a database of only spent fuel characteristics. Planned benchmarks for this system include the common materials test reactor (MTR) type research reactor and pressurized water reactor (PWR) power reactor samples and a complete set of results is expected by March 2015.
展开▼