An adaptive optimal scheduling and controller design is presented that attempts to improve the performance of beer membrane filtration over the ones currently obtained by operators. The research was performed as part of a large European research project called EU Cafe with the aim to investigate the potential of advanced modelling and control to improve the production and quality of food. Significant improvements are demonstrated in this paper through simulation experiments. Optimal scheduling and control comprises a mixed integer non-linear programming problem (MINLP). By making some suitable assumptions that are approximately satisfied in practice, we manage to significantly simplify the problem by turning it into an ordinary non-linear programming problem (NLP) for which solution methods are readily available. The adaptive part of our scheduler and controller performs model parameter adaptations. These are also obtained by solving associated NLP problems. During cleaning stages in between membrane filtrations enough time is available to solve the NLP problems. This allows for real-time implementation.