Distinct and Quantitative Validation Method for Predictive Process Modeling with Examples of Liquid-Liquid Extraction Processes of Complex Feed Mixtures

Abstract

As of today, industrial process development for liquid-liquid extraction and scale-up of extraction columns is based on an experimental procedure that requires tests in pilot-scale. This methodology consumes large amounts of material and time and the utilized scale-up equations are crude estimates including considerable safety margins. This approach is practical for well-known systems or low-value products coupled with high production scale, where such a scale-up methodology has less impact on the overall profitability. However, for new high-value products in biologics manufacturing, a process development based on process understanding and the use of validated process models is imperative. Therefore, a distinct and quantitative validation workflow for liquid-liquid extraction modeling is presented on the example of two complex feed mixtures. Monte-Carlo simulations based on the presented model parameter determination concept result for both examples in prediction accuracy comparable to the experiments and prediction precision within the deviation of the respective experiments. Identification of statistically significant parameters is demonstrated. The presented methodology for model validation will support the implementation of liquid-liquid extraction in the manufacturing of new high value biological products in regulated industries by providing a workflow to derive a Quality-by-Design compatible process model.