The study of Lovastatin production as a benchmark simulation model for bio-manufacturing processes

In this work an end to end bio-pharmaceutical process for the production of lovastatin is implemented into a Matlab/Simulink environment. The simulation is developed with the objective to be a generic reference model, as it captures the typical dynamics of pharmaceutical bio-manufacturing. It can then be used as an open-source test problem to study different process scenarios in dynamic mode. The objective of this work is to propose the benchmark simulation as a testbed for the development and comparison of different control strategies in dynamic mode. To this purpose a highly referenced control framework (the Skogestad method) is adapted and applied to the dynamics of the present model. The implemented framework includes a distributed control system (DCS) as regulatory control layer, and an MPC multiple-input multiple-output application as a supervisory control layer. As an alternative for the supervisory control layer, an override control framework based on process knowledge is developed and validated as a lighter solution. The different control setups are compared from a total productivity and stability point of view in a set of case-studies. As expected, the MPC application is the most stable solution, but in all the case studies the alternative override supervisory control gives comparable API final productivity. Pharmaceutical bio-manufacturing is a highly regulated industry in which the resources and time dedicated to process optimization and control are limited. The development of an end to end reference simulation with a limited number of unit operations is a step towards a smarter resource exploitation in the industry.