(6c) Modeling the Effect of Relative Humidity On Median Granule Size and Distribution Width

The pharmaceutical industry has increasing interest in achieving more robust, efficient and controlled processing. Granulation for instance is of one the key unit operations for the production of solid dosage forms such as tablets. Efficient granulation requires tight regulation of both median granule size and granule size distribution (henceforth known as distribution width). Amongst the different types of granulation techniques, fluid-bed granulation is one of the more frequently used techniques in the pharmaceutical industry. Fluid-bed granulation is a complex process with many inter-connected process variables. It is highly dependent on material properties, process parameters and equipment design. Specific to fluid-bed granulation processes, previous work has experimentally addressed the effect of relative humidity (RH) on median granule size [1]. Results show that increasing inlet air humidity resulted in greater median granule size.

In the present study, the effects of RH on both the median granule size (d50) and distribution width (d84/16) are studied by means of a population balance modelling approach. Typically, RH is an unmeasured disturbance that fluctuates depending on the RH of the ambient air. Hitherto, the effects of RH on various important granule properties have not been addressed in detail. In this study, RH is an explicit term that is incorporated in the model which is then used to simulate median granule size and distribution width. Results show that both d50 and d84/16 are sensitive to variations in RH at different processing conditions. Further more, model sensitivities are comparable to experimental sensitivities reported in [1]. As a result, RH can now be categorized as a measured disturbance which lends credence toward designing a feed-forward/feed-back control configuration as opposed to conventional feed-back control which would be the default control scheme given that previously, effects of RH could not be quantified in silico. Under regulatory and load changes, results show that overall control-loop performance is enhanced for a feed-forward/feed-back control scheme compared to a feed-back control scheme, that is essentially brought about by explicitly accounting for the effects of RH in a computer model.

References

1. T. Narvanen, T. Lipsanen, O. Antikainen, H. Raikkonen and J. Yliruusi, ?Controlling granule size by granulation liquid feed pulsing?, International Journal of Pharmaceutics, 357, 132-138, 2008.