Process Characterization

Identify Critical Parameters and Quality Attributes!

The influence of the different control parameters on the performance of a chromatographic separation is highly non-linear.

One possibility to test the process for robustness is to sample the whole design space and check if the desired performance can be obtained. Experimentally, this would be an extremely time and labor intensive endeavor. However, working in a model-based fashion allows to perform this sampling with an arbitrary resolution in the computer. The results can then be evaluated with standard methods of data analysis.

One the right, we built a partial least-squares (PLS) model from the results of a design space sampling of a periodic counter-current chromatography system.

With a hierarchical cluster analysis, four groups could be identified. Group 1 correlated with production rate, group 3 with yield and group 4 with purity. The contribution of the system parameters to these groups indicates the most critical ones for process performance. Based on this, only few lab experiments must be performed for a worst-case analysis.

Multi-criteria optimization result generated with ChromX Simulations generated and evaluated with ChromX.PLS model generated and evaluated with Simca by Umetrics Inc / MKS Instruments .