Multi-objective process optimisation of beer fermentation via dynamic simulation

Fermentation is an essential step in beer brewing: when yeast is added to hopped wort, sugars released from the grain during germination are fermented into ethanol and higher alcohols. To study, simulate and optimise the beer fermentation process, accurate models of the chemical system are required for dynamic simulation of key component concentrations. Since the entire beer production process is a highly complex series of chemical reactions with the presence of over 600 species, many of the specific interactions are not quantitatively understood, a comprehensive dynamic model is impractical.

This paper presents a computational implementation of a detailed model describing an industrial beer fermentation process, which is used to simulate published temperature manipulations and compare results with those obtained following the protocol currently in place at WEST Beer brewery (Glasgow, Scotland, UK). A trade-off between design objectives has been identified, making determination of a single optimal scenario challenging. A Simulated Annealing (SA) algorithm has been developed in order to pursue stochastic optimisation of the fermentor temperature manipulation profile, on the basis of generating an enormous set of plausible manipulations which adhere to suitable operability constraints at an appropriate level of temporal domain discretisation. The objective function considers ethanol maximisation as well as batch time minimisation (with variable weight allocation), and explicit constraints on diacetyl and ethyl acetate concentrations. Promising temperature manipulations have been determined, allowing for batch time reductions of as high as 15 hours: this represents a substantial decrease in production cycle time, and is thus expected to improve annual plant throughput and profitability, without any discernible effect on flavour.