Development and Application of an Open Access Bioreactor Performance Modelling Workflow for Cultivated Meat Yield Prediction and Optimisation

Abstract

The ability to predict the yield of cultivated meat production facilities is valuable in providing cost of goods sold predictions as part of techno-economic analyses and as a key performance metric in bio-manufacturing facility design and optimisation activities ahead of capital investment and facility construction. Existing cultivated meat techno-economic analyses either assume an achievable yield or predict it using model equations which are quick to solve but omit or simplify key performance characteristics. There is, at present, no openly available modelling workflow that enables yield prediction or optimisation for given bioreactor architecture, geometry and operational inputs. The present work builds on existing yield prediction approaches through the use of computational fluid dynamics to provide more targeted performance characterisation, coupled to a Bayesian optimisation workflow to optimise yield against bioreactor geometric and operational input parameters and made openly available. Application to 20,000 litre and 42,000 litre stirred tank reactors suggest bioreactor performance limited yield (not cell metabolism limited yield) could be higher than reported in previous studies. Application to a previously considered 300,000 litre air lift reactor design suggests that yields used in a recent techno-economic analysis are achievable, from a bioreactor performance perspective.