Michael Ghijs

BIOMATH, Model-based analysis and optimisation of (bio)processesMichael Ghijs
Department of Mathematical Modelling, Statistics and Bioinformatics

Coupure Links 653
B-9000 Gent (Belgium)
Tel.: +32-9-264.59.35
Fax: +32-9-264.62.20
E-mail: Michael.Ghijs@UGent.be
Education: Bio-engineer


Traditionally, production of pharmaceuticals is realized using batch processing steps which can be fairly inefficient, time-consuming and poorly upscalable. Recent clarification by the regulatory authorities as well as the industry have accelerated the search for new manufacturing technologies. Consequently, a transition towards continuous manufacturing which includes real-time process monitoring and closed-loop control, has gained tremendous attention over the past few years. This because of the economic, environmental, safety and product-related advantages of a continuous manufacturing line. Driven by this change in mind-set, an innovative continuous from powder-to-tablet line (ConsiGmaTM-25, ColletteTM, GEA Pharma Systems) was developed for secondary manufacturing of pharmaceutical tablets.

After wet granulation in the ConsiGmaTM-25, a drying step improves the processability and strength of the produced granules. A semi-continuous fluid bed dryer achieves this process step, in which a continuous inflow of wet granules is sequentially partitioned over six drying cells, in which parallel batch drying is performed. Part of the research objective consists of modelling drying and breakage in the system. Moreover, in describing the drying kinetics, attetntion is paid the the effect of granule size herein.

A second part of the research involves connection of the developed fluid bed dryer to a twin-screw granulation model in the framework of flowsheet modelling.

In order to describe fluid bed scale drying kinetics, the basics of single granule drying kinetics are modeled and applied to the semi-continuous fluid bed dryer system. For breakage a population balance model is applied. Both models are developed, calibrated and validated in Python.