Use of near-infrared spectroscopy for monitoring fermentation in a corn dry grind ethanol process

The objective of this study was to determine suitability of near-infrared (NIR) spectroscopy to predict various fermentation quality indicators such as ethanol, total soluble sugars, fermentable sugars, glucose, maltose, maltotriose, glycerol, and organic acids in the fermentation broth for a corn dry grind process. Effects of sample temperature on spectral absorption were identified and related to loadings of principal component scores. Partial least squares 1 (PLS1) calibration models were built on spectra obtained from a 5 mm pathlength transmission accessory and from a 0.5 mm transflectance accessory. Prediction results from calibration on both accessories were compared. Lactic acid and acetic acid were present at very low levels and did not appear to separate from the other fermentation constituents for the given pathlength and concentrations observed. External validation for ethanol had RMSEP values of 1.5% v/v, the larger portion of which was contributed by a large bias of 1.4% v/v, while the SEP (indicating precision of the model) was 0.6% v/v with two factors. External validation for total soluble sugars had RMSEP values of 1.5% v/v, the larger portion of which was contributed by the SEP with a value of 1.4% v/v, while the bias was 0.5% v/v with two factors. The RPD values obtained from the external validation for ethanol and total soluble sugars were 6.0 and 5.0, respectively. Total soluble sugar calibration had low RMSEP and high RPD values, while fermentable sugars, glucose, maltose, maltotriose, and glycerol had high RMSEP values and low RPD values. High RPD and low RMSEP values for both ethanol and total soluble sugar prediction validated the suitability of the near-infrared spectroscopy for monitoring of the fermentation process.