Near infra-red spectroscopic applications for milling: prospects and implications

Speed and precision of NIS data collection and recent developments in instrumentation offer potential for on-line analyses of product streams in sugar milling. This paper summarises the development of NIS analyses of components of clean fibrated cane, raw sugar, mill-prepared cane, and bagasse. Routine reflectance NIS was applied to these products, using sample preparation and/or presentation techniques developed for products with high moisture and/or heterogeneity. Calibration development primarily used PLS regression techniques on spectral data subjected to standard mathematical treatments. Excellent calibrations were developed for five major components of clean fibrated cane. Calibrations for ash, fines, moisture, and pol, for a population of Brand 1 and JA raw sugars, had SEP/SSD values <2.0, while those for colour, filterability and reducing sugars had values between 2.0 and 4.0. Calibrations for components of mill-prepared cane and bagasse were less satisfactory than clean cane calibrations. Small data sets and logistical problems with RLA data hindered calibration development. Excellent individual calibrations for dirt in cane suggested a robust application could be developed with a more substantial data set. As NIS analyses for mill-product streams are feasible, with appropriate hardware, process control applications of economic significance are a logical outcome.