Optimization of a granulation and tabletting process by sequential design and multivariate analysis

Sequential design and multivariate analysis were adopted to search for an optimal and robust granulation and tabletting process for a melt granulation process. The advantage with this procedure should be to produce tablets directly from granules prepared in one step without addition of any solvents. First the granulation process was investigated by a 26−2 fractional factorial design with six center points. The most important variables were identified, e.g., binder, PEG and HPMC. The main objective of the whole study was to investigate the influence of granules with different properties on the quality of the final tablet. Thus, five granulation batches, that spanned the experimental domain well in the first study, were chosen to vary along with four tabletting variables by a 25−1 fractional factorial design with seven center points. The results show that the properties of the granules are more important for the quality of the tablets than the settings of the tabletting machine. The center point was also found to be close to an optimal tablet performance. To confirm the robustness of the tabletting domain four experiments were run, including two noise variables (humidity and temperature) set at the extremes. The resulting PLS model showed very good predictability for most of the factors.