Quality evaluation of white sugar crystals using the friability test and their non-destructive prediction using near-infrared spectroscopy

This study quantitatively measures the degree of crystal fragility using near-infrared spectroscopy (NIR) with multivariate analysis to evaluate the formation of fine powder of sucrose crystals during transportation. The friability test was conducted on 50 batches of six grades of sugar crystal product samples. Subsequently, the loss of sample weight (LSW) was evaluated, and the NIR spectra (NIRS) of the samples were measured using an NIR spectrometer with a diffuse reflectance fiber-optic probe. The best-fitted calibration model to predict LSW was obtained based on NIRS without pretreatment using partial least squares regression (PLSR) method. In the result of the best-fitted model involving three latent variables, the correlation coefficient of the plot between the predicted and measured values was 0.702, indicating sufficient linearity. Comparative analysis of the regression and loading vectors of the calibration model suggested that the best-fitted calibration model for predicting LSW comprised the major physical and minor chemical information. The results of NIRS/PLSR and Kubelka-Munk methods suggested that physical and chemical mechanisms of the LSW increased the formation of fine powder of the sugar crystals during the friability test. The NIRS/PLSR method can be useful for real-time monitoring during the continuous manufacturing process of sucrose crystals. [Display omitted] •Destruction of sugar crystal in transportation was measured by the friability test.•Sugar crystal fragility could be evaluated as loss of sample weight.•Sugar fragility can predict by near-infrared spectroscopy with multivariate analysis.•PLSR calibration model to predict fragility was comprised three latent variables.•The prediction by NIRS/PLSR was consisted with that by Kubelka-Munk method.