Abstract
The purpose of this study was to quantitatively evaluate the water content of granules drying in a fluidised bed by near infrared (NIR) spectroscopy in real time and estimate a constant drying rate. Riboflavin granules were prepared by extrusion based on a standard formulation. NIR spectra collected during drying and the water content of granules was correlated by partial least squares (PLS) regression. To consider variability among batches, the leave-one-batch-out cross-validation procedure was performed. The PLS analysis showed that the plots of predicted vs actual values were linearly correlated with a coefficient content which can be continuously predicted with small errors. The difference between batches was clarified by results of score plots. In each batch, water content was predicted accurately using the PLS model. Based on water content values predicted in real time, constant drying rate was estimated. This result suggested constant drying rate might increase as the granule size decreases. This technique provides a robust calibration model for a better understanding and control of the drying process.
Keywords
Get full access to this article
View all access options for this article.
