Lautering Performance Prediction from Malt by Combining Whole Near-Infrared Spectral Information with Lautering Process Evaluation as Reference Values

Abstract

The aim of this study is to predict lautering performance from malt batches. It pursues an innovative approach by combining the total near-infrared (NIR) spectral information content with lautering process evaluation data. For this, multivariate data analysis is applied to combine the information content of whole NIR spectra of malt with lautering process data. Three NIR spectra were taken for each of the 11 single-malt batches of 100% pilsner malt. Additionally, standard laboratory analyses were performed. Standardized lautering processes were carried out, evaluated, and categorized in “good,” “semi,” and “bad.” The 51 lautering process evaluation values, each with three associated NIR spectra, add up to a total of 153 objects employed in building the prediction model. Partial least squares discriminant analysis (PLS-DA) was used for classification and calculating prediction accuracy. The prediction accuracy of the calibration set was 92.7% with only three failed lautering performance predictions. In the test set validation, only one lautering process prediction failed. This results in a prediction accuracy of 90.6% for the calibration model. With this study, the feasibility of NIR spectroscopy in combination with multivariate statistics to predict lautering performance by analysis of malt was proven.