(499a) Controllability Analysis and Identification of Optimal Control-Loop Pairings in a Multiple-Input Multiple-Output Granulation Process

Granulation is complex process whereby fine particles agglomerate to form larger granules, due to adhesive forces brought about by the liquid binder. Many of the existing granulation operations are in continuous mode and given the pharmaceutical industries' recent initiative to shift from batch to continuous, granulation in continuous operation assumes more importance [1]. In a continuous granulation process, feed material is continuously introduced into the granulator alongside the interplay of the various granule mechanisms such as nucleation, aggregation, consolidation and breakage. The granulator is fitted with several spray nozzles through which the liquid binder is introduced into the granule bed. The granules formed are then dried and classified based on product specification(s). Granules that do not conform to these specification(s) are recycled and reprocessed. Prior to the actual design and implementation of the controller for the granulation process, it is important to ascertain how well the process can be controlled and what factors may hinder the control-loop performance. It is also imperative that appropriate process inputs and outputs are selected for feedback control and that they are paired correctly as incorrect pairings may limit and hinder control-loop performance. A process is said to be controllable if there exists a controller that in principle, be able to achieve a certain output state via certain admissible input changes [2]. Controllability is an intrinsic process property and is determined by the process design such as sensor locations, recycle loops, internal couplings and sizing of equipment. Controllability conditions in granulation are also made more challenging given the presence of internal variables [3].

The aim of the present work is firstly to systematically ascertain the controllability of the granulation process by examining the effect of control handles such as binder and feed flowrates on controlled variables such as average size, distribution width, moisture content and bulk density. Secondly, to identify suitable control-loop pairings to facilitate a better understanding of the design and operation of the granulation process. Results are presented on a continuous and compartmentalized population balance model of a granulator, which depicted a realistic model of a pilot-plant granulation process. The model considered was based on granulation mechanisms that have been previously derived from first principles and validated at the batch-scale for different operating conditions and granulation systems [4,5]. Model simulations elegantly captured the nonlinear effect of nozzle positions on the granule output. As a direct consequence of the compartmentalization, the controllability of the granulation process could be examined, taking into account all the important process inputs and outputs, providing valuable insight into the control and operation of the granulation process. Results show that the average diameter, moisture content and bulk density are realistic choices of controlled variables given the previously identified manipulated variables at the operators' disposal. The size distribution width was not sensitive to any of the manipulated variables and deemed uncontrollable which also corroborates the experimental findings of Glaser et al. (2009) [6]. This is an important finding as size distribution is an important variable that needs to be tightly regulated. For instance, in the pharmaceutical industry, variations in granule size distribution (GSD) could lead to undesirable variations in tablet weight. As a result, this paves the way for alternative manipulative variables that have shown to influence the GSD as identified in the experimental work of Ramachandran et al. (2008) [7]. In this work, by means of step tests and a relative gain array analysis, several possible control-loop pairings were identified and within them, an optimal control-loop pairing was selected. Such a study on a MIMO granulation process lends valuable insight and knowledge which can potentially be applied to an actual granulation process.

References

1. C. Vervaet and J. P. Remon, ?Continuous granulation in the pharmaceutical industry?, Chemical Engineering Science, 60, 3949-3957, 2005.

2. R. A. Eek and O. H. Bosgra, ?Controllability of particulate processes in relation to sensor characteristics, Powder Technology, 108, 137-146, 2000.

3. D. Semino and W. H. Ray, ?Control of systems described by population balance equations ? i. controllability analysis?, Chemical Engineering Science, 50, 1805-1824, 1995.

4. J. Poon, R. Ramachandran, C. Sanders, T. Glaser, C. D. Immanuel, F. J. Doyle III, J. D. Litster, F. Stepanek, F. Y. Wang and I. T. Cameron, ?Experimental validation studies on a multi-dimensional and multi-scale population balance model of batch granulation?, Chemical Engineering Science, 64, 775-786, 2009.

5. R. Ramachandran, C. D. Immanuel, F. Stepanek, J. D. Litster and F. J. Doyle III, ?A mechanistic model for granule breakage in population balances of granulation: theoretical kernel development and experimental validation?, Chemical Engineering Research and Design, 87, 598-614, 2009.

6. T. Glaser, C. F. W. Sanders, F. Y. Wang, I. T. Cameron, J. D. Litster, J. Poon, R. Ramachandran, C. D. Immanuel and F. J. Doyle III, ?Model predictive control of continuous drum granulation?, Journal of Process Control, 19, 615-622, 2009.

7. R. Ramachandran, J. Poon, C. Sanders, T. Glaser, C. D. Immanuel, F. J. Doyle III, J. D. Litster, F. Stepanek, F. Y. Wang and I. T. Cameron, ?Experimental studies on distributions of granule size, binder content and porosity in batch drum granulation: Inferences on process modelling requirements and process sensitivities?, Powder Technology, 188, 89-101, 2008.