Prediction of Confidence Limits for Diacetyl Concentration during Beer Fermentation

Abstract

A dynamic model for diacetyl production and reduction was developed based on experimental data from 14 laboratory-scale (15-L) lager beer fermentations carried out in various conditions of temperature (10–16°C), top pressure (50–850 mbar), initial yeast concentration (5–20 million cells per milliliter) and initial wort gravity (1,036–1,099 g/L). Uncertainties due to measurement errors, model parameters, and batch-to-batch variability were described in a probabilistic framework. The model predicts a probability distribution for the final diacetyl concentration from which a median value and an upper boundary, at a specified confidence level, are derived. It is demonstrated that in-line diacetyl measurements at early stages of fermentation greatly reduce the uncertainty about the final diacetyl level in each specific batch.