Real time quantification of carbohydrates in beer wort by FTIR and neural network analysis

The brewing process of wort has been predominantly empirical, once that just after the production of a batch is possible to evaluate the performance of raw materials, and if its necessary makes adjustments for the next batches. In this context, the real time information about the composition of the beer wort can facilitate the implemention of modifications during the process to improve both efficiency and precision of the desired physicochemical profile. Thus, this work shows the development of an equipment and signal processing that combines three techniques for monitoring the mashing process in beer production. Real-time quantification of carbohydrates produced by enzymatic reactions between barley malt enzymes and starch during the saccharification step, which is usually achieved through spectral analysis using Fourier transform infrared absorption in the mid-infrared region through collection and referring samples in real time. The system incorporates principal component analysis and artificial neural networks for data processing, however by implementing a flow measurement system controlled by software, the concentration of major carbohydrates (glucose, maltose, maltotriose, and dextrins) in the wort can be quantified during the enzymatic reactions in real time. These measurements offer new possibilities for process control, error reduction, and optimization of reactions as needed, allowing fine control of the process. The real time spectroscopy will provide the application of machine learning tools in the process. While the device and procedure is demonstrated for the efficient control of the beer production, its context and applicability goes much beyond such applications.

The online version contains supplementary material available at https://doi.org/10.1007/s13197-024-06128-4.