Modeling For Chromatographic Separation And It's Application In Pharmaceutical Purification Process

Chromatography is an efficient separation technique, which is widely used in chemical and pharmaceutical engineering. In this study, modeling for the non-linear chromatography was studied in detail, as well as its application for chromatographic simulation and optimization.Chromatographic separation of the two nucleotides CMP and UMP was simulated by the Equilibrium-Dispersive (ED) model in this study. The adsorption isotherms in the ED model were determined by the inverse method. Prediction performance of the model was tested and quantitatively compared under three different kinds of conditions. The results shown that prediction performance of the ED model was very sensitive to different selection of adsorption isotherm, and ED model with bi-Langmuir isotherm seems especially suitable for simulating chromatographic separation of CMP and UMP, rather than that with Langmuir or modified Langmuir isotherm.In order to improve precision of prediction, the GR model was used for simulation of chromatographic separation of Naringin and Neohesperidin. Differential equations of the GR model were numerically solved by the Orthogonal Collocation on Finite Elements (OCFE). The number of finite elements, collocation points in bulk and particle phase were optimized for the GR model of Naringin and Neohesperidin in this part. And the result of simulation indicated that the performance of GR model was much better than that of ED model.Application of the model in simulation of gradient elution, optimization and amplification of chromatographic separation of salviamolic acid A and B were further discussed. And the modified GR model described the gradient elution of two-component system quite well. Then the ED model was used for optimizing the injection volume and flow rate, which were two significant operational parameters, and the result of optimization was then experimentally validated. Finally for some amplified condition, the computer-aided optimization was done and compared with experiential estimation.