Process system modelling of TSG
From a process technology perspective, TSG are often divided into multiple zones arranged in series, linking each granulation step to the next (Figure 2). Changes in screw configuration (number and location of transport and kneading element) lead to changes in the dominance of one granulation mechanism over the other causing different granule size distribution (GSD) and granule properties. While the screw modules have preference for individual rate processes due to design characteristics, the operational regimes within TSG length are not completely decoupled due to inherent needs of granulation processes. This prompts for substantial process understanding both at particle and system (containing barrel) levels, and requires a multiscale approach. Applications of population balance modelling (PBM) (system level) and discrete element method (DEM) (particle level) approaches in granulation have shown their relevance in modelling batch granulators and mixers in past. Hence, the opportunity exists to adapt these modelling approaches for appropriate numerical analysis of TSG. However, this adaptation requires consideration of material and equipment properties along with a comprehensive list of process variables and status (Figure 3). These basic models at different levels should be linked using multi-scale integration frameworks in such a way that the granule scale model supply the agglomeration kernel to the system scale model. To do so, the granule scale model requires the current GSD and the volumetric hold-up of the granules from the barrel scale. This approach has provided good results in other studies with continuous fluidized bed granulators . Also, because the TSG are not well-mixed system like HSM, to include system heterogeneity in the TSG it is required to combine DEM and PBM via a compartmental model.
Tools needed for real-time process measurement
There has been a significant development in the measurement techniques which have been used in granulation studies using TSG. These developments focused on measurement of parameters 1 which are either quality parameters or indirectly used to determine the quality of granules as discharged product [2,3]. However, for mechanistic understanding and validation of rheo-kinetic models, local information about numerous parameters such as granulation liquid, filling degree of the barrel and many more mentioned in Figure 3 are needed. Several crucial process parameters which cannot be easily measured are correlated with in-line dynamic torque which is 0-dimensional measurement, making it not suitable to provide local information. Recent studies apply radioactive particle tracking methods such as Positron Emission Particle Tracking (PEPT), and imaging techniques such as particle image velocimetry (PIV) for flow visualization in a barrel in-line. The obtained velocity profile in TSG further utilized to construct residence time distribution (RTD) profiles and study the effect of a change in viscosity of the granulation liquid . Measurement techniques, for instance, magnetic resonance imaging (MRI) which are capable of examining processes non-invasively and already used in other areas of research, also facing the challenge of opaque multiphase systems, needs to be investigated. Considerable attention has been paid to development of several rapid and non-destructive on-line soft-sensing methods to estimate hard-to-measure on-line quantities through chemometric models. The capability and applications of near infra-red and Raman spectroscopy to provide both chemical as well as physical information such as moisture content and particle size on a real-time basis using chemometric methods has been utilised in various measurements. Soft sensors based on partial least squares (PLS) regression or principal component analysis (PCA) are often preferred, since these methods are well-known in the pharmaceutical industry which facilitates validation .
TSG are generally operated continuously, thus the focus of modelling so far is on steady-state processes for process study. However, in addition to the spatial model dimensions, time may be a key factor in TSG which requires deeper insight. Also, the application of DEM or computational fluid dynamics (CFD) simulation which can provides particle tracking information have greater role to play. Besides the computational data, measurements from experiments would require more sophisticated in-line measurement tools in future, for validation before the simulated profile can be applied with confidence in practice. Despite the scientific challenges, the development of modelling techniques and robust measurement tools
 N N Rao. Simulations for modelling of population balance equations of particulate processes using the discrete particle model (DPM). PhD thesis, Otto-von-Guericke-Universität Magdeburg, Universitätsbibliothek, 2009.
 Fonteyne, M., Vercruysse, J., Díaz, D., Gildemyn, D., Vervaet, C., Remon, J., & Beer, T. (2013). Real-time assessment of critical quality attributes of a continuous granulation process Pharmaceutical Development and Technology, 18 (1), 85-97 DOI: 10.3109/10837450.2011.627869
 Vercruysse, J., Córdoba Díaz, D., Peeters, E., Fonteyne, M., Delaet, U., Van Assche, I., De Beer, T., Remon, J., & Vervaet, C. (2012). Continuous twin screw granulation: Influence of process variables on granule and tablet quality European Journal of Pharmaceutics and Biopharmaceutics, 82 (1), 205-211 DOI: 10.1016/j.ejpb.2012.05.010
 Dhenge, R., Washino, K., Cartwright, J., Hounslow, M., & Salman, A. (2013). Twin screw granulation using conveying screws: Effects of viscosity of granulation liquids and flow of powders Powder Technology, 238, 77-90 DOI: 10.1016/j.powtec.2012.05.045
 Gernaey, K., & Gani, R. (2010). A model-based systems approach to pharmaceutical product-process design and analysis Chemical Engineering Science, 65 (21), 5757-5769 DOI: 10.1016/j.ces.2010.05.003
The research groups of Prof. Thomas De Beer at Laboratory of Process Analytical Technology, Ghent University in Belgium focuses on the implementation of PAT systems in innovative pharmaceutical production processes and works therefore in close collaboration with the Laboratory of Pharmaceutical Technology (Prof. J.P. Remon and Prof. C. Vervaet) and is part of the QbD and PAT Sciences Network. The study presented here an abridged version of project under-progress in collaboration between the PAT research group (Faculty of Pharmaceutical Sciences, Ghent University, Professor Dr. Thomas De Beer), the Process Engineering and Technology research group (Technical University of Denmark, Prof. Krist Gernaey) and the BIOMATH research group (Faculty of Bioscience Engineering, Ghent University, Prof. Ingmar Nopens).