Dynamic Modeling of a Liquid-Liquid Extraction Column-An Industrial Approach

Model-based evaluation and optimization of technical extraction columns are necessary when the product specifications of a technical system (TS) are not reached. However, state-of-the-art models are parameterized and validated using standard test systems (STSs) of highest purity. In contrast to that, TSs are characterized by an incomplete list of components, impurities, and batch-depending physicochemical properties. This divergence between STSs and TSs results in a lack of trust regarding developed model-based approaches for extraction columns. In this paper, an experimental approach for parameterization of a one-dimensional rate-based model is presented. The model is parameterized for a STS leading to a weighted absolute percentage error, ((sic))WAPE(wAcetone,out), of 8.53%. The parameter set is then used for a TS leading to a ((sic))WAPE(wTC,out) of 17.43%. The distribution coefficient is the most sensitive model parameter as a change in the distribution coefficient of 3% yields a relative change of ((sic))WAPE(wTC,out) by 19.39%. Varying the fluid dynamics parameters Sauter diameter and hold-up by 20%, the relative change in ((sic))WAPE(wTC,out) is 44.03% and only 5.48%, respectively. A recommendation for minimal instrumentation for lab-scale extraction columns and technical large-scale columns is given.