ABSTRACTThe nuclear conversion plant operated at Necsa in the past resulted in the formation of a large amount of waste material containing ~ 50% U, which has reuse potential if the recovery thereof can be achieved. The waste material contains various impurities of which F, Ca, and Fe have been identified as the main contributing impurities after dissolution. These impurities are present in the waste material in varying concentrations and could interfere with the U recovery. During this investigation, the possibility of varying impurities being present during purification was addressed by simulating the waste material using synthetic solutions to investigate the effect of different concentrations of the main impurities, as well as nitric acid and TBP concentrations on the extraction of U. The results from this simulated approach will serve as motivation for obtaining more of the real waste samples for further investigation and confirmation of results. This was done using a statistical design of experiments approach, giving a predictive model which can be used to predict U recovery for a given waste sample composition. It was found that nitric acid had the most significant influence on U extraction, where an optimum amount (97%) was achieved in 5 M nitric acid at, or above 30% TBP. At low nitric acid concentrations, it was found that the U extraction increased with increasing TBP and decreasing F concentrations. An unexpected interaction between F and Fe was observed, which led to an increase in U extraction as the Fe concentrations were increased due to complex formation with F, which limits the amount of F available which would have decreased U extraction due to competition. Furthermore, a mathematical model was derived which describes the U extraction as a function of varying impurity, nitric acid and TBP concentrations. Optimum extraction conditions were confirmed using a feed solution containing additional impurities (Al, Na, K and Si) which are present in the waste material in concentrations above 0.5 wt %. While confirming the optimum extraction conditions, the validity of the statistical model was also confirmed from these experiments, indicating its suitability for U extraction predictions as a function of impurity concentrations.