Representative Sampling and Radiological Characterization of Heterogeneous Legacy Waste - 22031

A major challenge facing many nuclear operators is characterization of legacy wastes with different activity levels, invariably containing mostly heterogeneous bulk materials, e.g. concrete and soil. This endeavor is a very complex task, involving the ability to make reliable categorization of the material per treated unit (drum, container, ... ) based on the relevant waste acceptance criteria (WAC) as well as paying particular attention to the long-term safety criteria. It is a fact that some of the most critical radionuclides considering long-term safety are difficult-to-measure (DTM), meaning that non-destructive assay (NDA) is unsuitable to perform these tasks. A commonly used approach is the combination of NDA and scaling factors - based on correlations between DTM radionuclides and easy-to-measure (ETM) or key radionuclides. However, this approach is often associated with large uncertainties due the tacit or explicit assumption of 'homogeneous distributions' of both content and radioactivity. In this project, these uncertainties are expected to be too large to allow direct implementation of this methodology for a range of critical parameters. Because of the historical contents and the low concentration limits in the WAC - both for conditional release and low-level radioactive waste disposal - a characterization process was developed based on 100% visual inspection and sorting followed by physical sampling. The material to be characterized is transferred to a low-activity hot-cell where a thorough visual inspection is conducted manually followed by guaranteed representative sampling of all material batches involved. A robust representative sampling process was developed based on the Theory of Sampling (TOS). A so-called Replication Experiment was conducted to survey and quantify the intrinsic heterogeneity of selected typical materials for a range of critical parameters. Despite the results indicating that the radiological contamination is very heterogeneously distributed, a primary sampling process was developed with a significantly better accuracy then commonly used NDA techniques. The overall experience gained in the first year of this project is discussed and some future challenges are highlighted. (authors)