A Neural Network Tool for Brewery Fermentations

Fermentation is an important, if not the most important process step in the production of beer. It is subject to alterations stemming from the variation in the yeast, a living organism, and due to the complex raw materials of biological origin. The variability may manifest in changes in the fermentation speed or cause off-flavors in beer. Thus, there is a need for tools, both analytical and computational, to help in monitoring the process and keeping it in a desired course. In this paper we describe a prediction tool to assist production management in the brewery. The system relies on a neural network that predicts the course of the fermentation based on yeast history, fermentation recipe and raw material variables. The system is able to predict the fermentation time with an average error of 10 percent of the remaining fermentation time, which enables the brewery to spot problematic batches days in advance and plan the correct yeast harvesting time with an increased accuracy. The system has been in daily trial use in Hartwall’s brewery in Lahti since March 2000 and the user experiences have been positive.