Statistical Modelling for Optimisation of Mash Separation Efficiency in Industrial Beer Production

Mash separation is a critical pre-processing step in beer production, ensuring that a high-quality stream of solubilised grain carbohydrates and nutrients (wort) is fed to the fermentors, in which sugars are then biochemically converted to ethanol. This essential pre-fermentation step is performed via either of two key units (lauter tuns or mash filters); the output quality of the clarified liquid stream (wort) depends on numerous critical variables (grain composition and size distribution, mash mixture physicochem. properties, brewing recipe, separation conditions). While first-principles mathematical descriptions may remain elusive, a multitude of available (input-output) industrial data can be used to improve understanding. This paper explores causality via statistical (Partial Least Squares) models for two types of beer, and performs a sensitivity analysis using the proposed input-output correlations towards mash separation improvements. Strong wort volume and incoming feed quality to the mash filter emerge as having the strongest effect on filtration time, a key industrial performance metric for optimisation.