Katsoulas, Konstantinos; (2025) Model-based approaches for accurate and efficient method development in liquid chromatography. Doctoral thesis (Ph.D), UCL (University College London).

Text Konstantinos _Katsoulas_Thesis.pdf - Submitted Version Access restricted to UCL open access staff until 1 April 2026. Download (24MB)

Abstract

Pharmaceutical production is often hindered by the downstream processing, which involves recovering, purifying, and concentrating active pharmaceutical ingredients (APIs). Chromatography plays a key role in API purification. With increasing computational power, model-based optimisation has emerged as a powerful tool to enhance chromatographic efficiency, reducing costs and time from lab-to-market while maintaining high product quality. Model-based process optimisation relies heavily on how accurately the models represent the underlying physics of the process. Developing such representative models, however, is costly and time-consuming. This thesis clarifies the widely used equilibrium dispersion model (EDM) by deriving it through rigorous asymptotic analysis, thereby identifying its correct expression. Additionally, a mathematical framework is established that justifies replacing the more general pore diffusion model (POR) with the simpler EDM, thereby promoting the use of a simpler model when possible and strengthening the theoretical foundation of the modelling efforts. Chromatographic models rely on isotherm equations to capture the quasi-equilibrium between mobile and stationary phases. Selecting the right isotherm model and estimating its parameters require extensive experimentation and process intuition. To address this, a step-by-step isotherm model identification methodology was developed, leveraging model-based design of experiments (MBDoE) and statistical tests, thereby reducing experimental effort and promoting more informed decision-making. Additionally, a diagnostic procedure is proposed to refine underperforming isotherm models, improving their structure and accuracy in representing experimental data. By reducing reliance on conventional experiments (e.g. factorial), this methodology will enhance efficiency but also make model development more accessible to researchers and industry practitioners. As automation and artificial intelligence methodologies continue to advance, the principles established in this work can serve as a foundation for fully autonomous isotherm model identification, paving the way for smarter and more adaptive pharmaceutical manufacturing processes.

Download activity - last month

Export as XMLJSONCSV

Download activity - last 12 months

Export as JSONXMLCSV

Downloads by country - last 12 months

Export as JSONXMLCSV

Archive Staff Only

View Item