Reliable estimation of the key variables and of their rates of change in alcoholic fermentation

The paper establishes a rigorous probabilistic framework for the reconciliation of apparently conflicting data from various physical and chemical measurements related to the key biological variables of alcoholic fermentation: the ethanol and the residual sugar concentrations. The analysis is carried out on a database consisting of 15 beer fermentation experiments, for which off-line determinations of ethanol concentration, fermentable sugar concentration, wort density and refractive index are available, as well as on-line records of evolved CO₂. The basic reconciliation method uses mass balance and monotonicity constraints derived from the biological knowledge of the fermentation process. In order to provide interpolated values and rate estimates, smoothness requirements are added. The reconciliation procedure gives more reliable estimates than any given measurement, detects outliers, helps fixing problems in the experimental setting and is also applicable on line.