MODELING OF RARE EARTH SOLVENT EXTRACTION PROCESS FOR FLOWSHEET DESIGN AND OPTIMIZATION

The separation and purification of rare earth elements (REEs) into individual products has been a topic of significant interest for researchers and engineers for many decades. The prime reason for such sustained interest is due to REEs’ demand and application in modern technology, as well as the challenges associated with their separation and purification. The chemical similarity of rare earth group elements is responsible for difficult separability which makes purification of individual elements challenging. Despite associated complications, processes such as solvent extraction (SX) and ion-exchange have been successfully utilized in the separation and production of REEs on pilot and commercial scales. Of the two-processes, SX is popular because of its capacity, continuous nature, fast reaction kinetics, and ease of operability. However, the literature and work on SX process design and flowsheet development for the separation of REEs is scarce. Previous studies on the separation of REEs using SX has been focused on the experimental aspect of improving separation factors by the use of new extractants or combination of extractants. However, in a continuous SX process, the separation effects are transformed because of the multi-stage nature of operation and intricate interaction between variables, making the process more complicated and difficult to design. To have both a qualitative and quantitative assessment of such a complex process is challenging. For a rare earth system, complication compounds because of the diverse feed nature depending upon the source, multiple elements and the proportion of the individual elements present in the feed. Separations for such systems using SX require numerous stages, the determination of which is not well established and traditional methods such as McCabe Thiele becomes impractical to use because of the multiplicity of similar extracting elements. Designing and testing such processes on a pilot or industrial scale is not only time consuming but also cost and labor intensive. This work provides a novel design framework utilizing equilibrium analysis of the rare earth SX process combined with a process modeling methodology in a modular framework to design a flowsheet for REE separation. The use of process modeling as an alternate to conventional McCabe Thiele allows analysis of a complex multi-component integrated SX system holistically. The approach is applicable to any feed composition and metal separation using SX. The equilibrium analysis for this study involved experimentally determining the separation of elements at different equilibrium pH and phase ratios. The experimental work was performed on a mixed rare earth salt solution containing yttrium, gadolinium, samarium, praseodymium, neodymium, cerium, and lanthanum. The distribution of elements was derived from a rare earth oxide product obtained from a coal-based source, part of ongoing research at the University of Kentucky. A DEHPA and TBP mixture was used as an extractant. Similarly, stripping experiments were carried out on loaded organic at different equilibrium acid molarities and phase ratios. The results obtained from the experiments were utilized in developing non-linear separation models. The models were integrated in a process- modeling framework and programmed in Matlab/Simulink as modular function blocks to describe loading, scrubbing and stripping processes involved in a SX operation. The blocks were then arranged and interconnected to design and simulate a multi-train SX flowsheet for individual or group separation of elements. A particle swarm optimization algorithm was applied to determine stage combination, resulting in the best separation of elements based on a defined objective function using recovery and purity of elements. Simulation and optimization showed good separation for yttrium and lanthanum from the feed mixture to a purity of 99.52 and 85.41, requiring 8-12-3 and 10-3-5 loading-scrubbing-stripping stages, respectively. Simulation results also indicated moderately difficult separability between gadolinium and samarium, and difficult separability for praseodymium, neodymium, and cerium groups