Automated Simulation and Parameter Estimation of Chromatography Processes

Sammanfattning: Preparative chromatography is an important separation technique in downstream processing for several industries. It has the ability to separate two or more component from each other to achieve high purity. Most components that are separated with preparative chromatography are high value components and it is therefore necessary to use small amounts in the development of the process too lower costs. The ability to model these systems and simulate them before running experiments could lower these costs and also make the process more efficient. At the Department of Chemical Engineering at Lund University, liquid chromatography systems (ÄKTA from GE) are controlled with a in-house developed software called \emph{Orbit}. In this thesis, a simulation software was integrated into the software so that an experienced user could simulate methods with the same syntax as before. The software was able to simulate the whole flow path with tubes, columns and other units. The user was then able to follow different components throughout the system and predict their behavior. For the column, a simplified mass steric action model was implemented so that the adsorption of components could be described. A parameter estimation software was also implemented into the software package. This gave the user the option to estimate model parameters for a component using only two data sets. The software designed so that \emph{Orbit} could conduct the experiments on the ÄKTA machines and then from that data, it could estimate model parameters. In this study, data sets was created using the simulation software and known model parameters. Using these data sets, the software was able to estimate the model parameters in different cases with great precision and robustness. Even if the user made a bad guess, the software was still able to find precise parameters.