CFD modeling of the turbulent precipitation of plutonium oxalate in a vortex reactor

Full text of publication follows:The nuclear fuel reprocessing, as it is done in La Hague COGEMA's plant, involves a precipitation in an un-baffled stirred tank to turn plutonium nitrate into plutonium oxalate prior to a calcination for using it in MOX fuel (Mixed Oxides). In every crystallization and precipitation (reactive crystallization), the key variable is sur-saturation, resulting from either a chemical reaction between two liquids, a liquid and a solid, a gas and a solid, or a decrease in the product solubility induced by temperature gradient or mixing of a solvent and an anti-solvent. According to the local sur-saturation degree, different mechanisms may occur leading to the solid dispersed phase: nucleation, growth and eventually aggregation, breakage and ripening. The challenge in that kind of modeling lies in the non linear behavior of the physical and chemical phenomenons, and in the different time and length scales involved which can not be solved without resorting to computational fluid dynamics (CFD). The modeling philosophy used here is divided in two parts: (1) a lower mesh statistical mixing model (FM-PDF) and a liquid energy spectrum model coupled to the RANS equations, for taking into account the continuous liquid phase mixing at various levels: reactor scale, turbulent dispersion and molecular diffusion; (2) a population balance model for the solid dispersed phase to link flow pattern to the morphology and chemical properties of particles. Many approaches have been developed so far to solve PBE but the recent breakthrough that allowed implementation of that equation in CFD codes is QMOM: quadrature method of moments, that represents the particle size distribution (PSD) in a finite number of delta functions, corresponding to abscissas and weights of the quadrature, which are used to calculate the mean particle size or whatever moment of the PSD. Part (1) has been successfully validated experimentally by means of an acid-base neutralization reaction; part (2), i.e. the PBE modeling coupled to experimental nucleation and growth laws and an aggregation kernel based on local shear rate, has given satisfactory results, strengthening the promising abilities of this local approach. (authors)