Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
1181564 | Chemometrics and Intelligent Laboratory Systems | 2009 | 6 Pages |
Fluidized bed granulation of a pharmaceutical mixture and wet granulation in general is an unpredictable and complex process. Usually the few variables that are known to indicate the granulation end-point are monitored independently but at the same time simultaneously. If one wishes to control the complex process and evaluate how this is affected by different physical and chemical granulation phases and variables measured then multivariate analysis and appropriate controlling tools need to be incorporated into the process line. It is also important to obtain early warnings from the process line when the process is already or about to develop in an unwanted direction. Multivariate tools in conjunction with process variables or spectroscopic data can reveal the latent structure of the process. In this study, multi-way models together with a few process variables have been utilized to distinguish successful batch granulations from unsuccessful runs. The batch-to-batch variation of different runs is handled using the special multi-way data structure of the experiments.