The Rokkasho Reprocessing Plant, located in northern Japan, includes facilities for reprocessing spent nuclear fuels and immobilizing high-level waste (HLW) into a stable-glass waste form based on borosilicate. The vitrification process consists of two large liquid-fed, joule-heated, ceramic-lined melters (LFCMs). During the active test campaigns at the Rokkasho Vitrification Facility, unstable melting performance and difficulty in glass operation were observed in the LFCM operation. The operating protocols that had been developed in the earlier full-scale mock-up testing with inactive waste simulants were found to be inadequate to manage a stable LFCM operation.

We needed to understand correctly the fundamental causes for the operational difficulties experienced in the active test. First, we studied and summarized the troublesome element behaviors and disturbances in the HLW vitrification process, primarily on the LFCM. Second, we proposed an empirical model, including the transition of the phenomena that appeared in the active LFCM operation. Third, we analyzed the operating data and the trend of operation indexes to find out the fundamental causes for the unstable LFCM operation based on the results obtained in the described model.

It was revealed that the presence of a significant amount of molybdenum and sulfur, together with the noble metals (NMs) Ru, Rh, and Pd in the Rokkasho HLW stream could lead to the formation of molybdate and sulfate salt phases [yellow phases (YPs)], and subsequently, the sedimentation of NMs at the bottom of the melter. The YPs may concentrate at the melter bottom and clog the pouring nozzle prior to the bottom accumulation of NM sludge, which can disrupt power deposition and cause difficulties during glass pouring operation.

Based on these investigated results, we provide insights into short-term countermeasures, including numerical control of the heat balance in the melter, dilution of HLW feed with inactive waste simulants, and periodic rinsing of the bottom glass. We also provide an evaluation of their validities on the active operation through full-scale mock-up testing. However, the periodic rinsing and dilution operation produce additional glass-filled canisters and increase the total backend cost.

Finally, we see a couple of important directions for future research and development. An advanced LFCM system has been developed for the long-term countermeasures to introduce recent advances in the LFCM design with YP and NM compatibilities. Furthermore, the modified or new glass formulations, which can incorporate more waste oxides into the glass matrix, known as high-waste loading glasses, also have been developed in parallel.