High performance finite element methods for three-dimensional chromatography models

Více o knize

Simple one-dimensional models are commonly used for numerical simulations in column liquid chromatography, regardless of column size. However, in microscale columns, local inhomogeneities in packing structure, flow fields, and concentration profiles can significantly impact the separation process. To analyze these effects and identify optimization potential, highly resolving three-dimensional simulations are necessary. This thesis presents a novel and efficient simulation strategy for spatially resolved chromatography models, focusing on a finite element solver tailored for fluid flow and mass transfer in complex geometries. A space-time Galerkin/least squares finite element method is employed to achieve stable solutions for non-linear and advection-dominant problems. The finite element mesh is partitioned effectively, enabling the solution of realistic problems on thousands of compute cores. Selected test problems demonstrate the simulator's capability to handle realistic scenarios in column liquid chromatography. However, for large problems, numerical approximation errors due to reduced mesh resolution may need attention. The solver is used to investigate differences in loading behavior between mono- and polydispersely packed microscale model columns. These simulations reveal that factors like column confinement and particle size, often overlooked in conventional chromatography models, can significantly influence microscale