(502a) Scale-up Modelling and Analysis of a Pharmaceutical Crystallization Process
This work presents an approach to develop a predictive model of a pharmaceutical crystallization process scale-up based on compartmental modelling. While full blown CFD models could be used to better describe and predict mixing characteristics in reactors as opposed to compartmental modeling, they are not practical for industrial applications due to heavy computational burden. Compartmental modelling is such that a trade-off approach is used to overcome the limitations of well-mixed models, by considering localized mixing, heat transfer and fluid hydrodynamics separately from crystallization kinetics within a crystallizer. While compartmental modelling is a well-known and successfully applied technique in other areas of process engineering, such as characterizing mass transfer for oxygen in industrial fermentation, there is a lack of comprehensive studies on the application of compartmental modeling approach to the scale-up of industrial pharmaceutical crystallization.
In order to apply compartmental modelling to pharmaceutical batch cooling crystallization, the crystallizer is divided into finite number of compartment volumes in a way that every compartmental zone possesses minimized or negligible gradients in profiles, such as crystal distribution, super-saturation and energy dissipation within their individual volumes. The same set of model equations and model parameters are defined in order to solve the balance equations for the conservation of population, mass and energy for every compartment. However, spatial variations in the crystallizer are modelled by different rates of nucleation, growth, dissolution and attrition in the individual compartments. The compartment modelling is implemented in MATLAB/Simulink.
The process behavior of the pharmaceutical batch cooling crystallization is analyzed with respect to crystallizer scale, process variables and operation conditions. The influence of the non-uniformly distribution of the related process variables is shown by comparing the multi-compartment model with a single-compartment well-mixed model.
We would like to thank the Danish Council for Independent Research (DFF) for financing the project with grant ID: DFF-6111600077B.