On the Conservatisms of Nepheline Predictors for Canister Centerline Cooled (CCC) High‐Level Waste (HLW) Glasses

ABSTRACT

Multiple models and feature sets have been generated, which attempt to predict the crystallization of nepheline (NaAlSiO ₄ ) in high‐level radioactive waste (HLW) glasses under canister centerline cooling (CCC) conditions. Earlier models were found to be overly conservative, while newer models had a not insignificant rate of false positives and negatives, with a drop off in predictive accuracy along the nepheline prediction boundary. Only partial or limited explanations for this drop in accuracy have hitherto been put forward, for example, experimental inaccuracies or insufficient data points. This research presents evidence consistent with a primary cause being differences in surface area to volume ratio (SA/V) between small experimental glass samples and large‐scale canister glasses. Furthermore, a new simplified, simulated (non‐radioactive) HLW glass composition (KE1) has been designed as a starting point for experiments that focus on the low accuracy region of nepheline predictor models, lying between compositions that are prone to precipitate nepheline and compositions that are not. This glass has been characterized using multiple techniques, and several experiments have been designed and implemented to highlight the marked effects of SA/V on the outcome of the product consistency test (PCT‐B, ASTM C1285‐21) used to validate the chemical durability (as measured by PCT tests) of nuclear waste glasses destined for geological disposal.