UC Irvine

Applying spectroscopic tools for chemical processes has been intensively studied in various industries owing to its rapid and non-destructively analysis for detecting chemical components and physical characteristic in a process stream. The general complexity of separation processes for used nuclear fuel, e.g., chemical speciation, temperature variations, and prominent process security and safety concerns, require a well-secured monitoring system to provide precise information of the process streams at real time without interference. Multivariate analysis accompanied with spectral measurements is a powerful statistic tool that can be used to monitor this complex chemical system. In this thesis, chemometric models that respond to the chemical components in the samples were calibrated and validated to establish an inline near real time monitoring system. The models show good prediction accuracy using partial least square regression analysis on the spectral data obtained from NIR, Raman and UV/Vis spectroscopies. Ultimately, an extraction process using a single stage centrifugal contactor was tested in our laboratory to determine the performance of an inline near real time monitoring system for a solvent extraction process representative of used nuclear fuel separation processes.