PhD Defence by Simon Brædder Lindahl

Principal supervisor:
Professor Gürkan Sin, DTU Chemical Engineering

Co-supervisors:
Professor Krist V. Gernaey, DTU Chemical Engineering
Dr. Deenesh Kavi Babi, Novo Nordisk, Denmark
Director Marianne Tolstrup Langfrits , Novo Nordisk, Denmark

Examiners:
Professor John Bagterp Jørgensen, DTU Compute
Professor Ana Barbosa Póvoa, University of Lisbon, Portugal
Professor Pierre Pinson, Imperial College London, United Kingdom

Chairperson at defence:
Senior Researcher Merlin Alvarado Morales, DTU Chemical Engineering

Popular Science Summary

Pharmaceutical products improve and save the lives of human beings around the planet. Production of the active pharmaceutical ingredient (API) takes place in a multi-stage system composed of manufacturing lines that are shared among a set of products. Such complexity imposes planning challenges related to the utilization and expansion of the installed system capacity. Production planning must consider resource sharing and campaign-mode production while capacity planning must determine the location, size, and timing of capacity expansions. In addition, manufacturing shutdowns for expansion project implementation must be tightly coordinated with production planning to maintain a stable supply of products and intermediates.

This PhD project focuses on the development of methods and optimization models to support decision-making on capacity planning, production planning and their integration with a specific focus on API manufacturing. A framework is presented to tackle general problems in manufacturing systems through standard system representations, mathematical models, and a systematic method. The 9-step method starts with a problem definition and represents the physical system as a directed graph of materials and manufacturing tasks. The graph is then translated into a mathematical model of the physical system and the model is solved in a series of solution steps. The solution complexity is increased along the steps starting from a production planning problem and ending with an integrated capacity and production planning problem including line retrofit and construction under uncertainty. Novel MILP models have been developed to consider mid-term planning through production campaigns and long-term planning through the aggregated allocation of manufacturing resources.

The framework has been applied to various case studies inspired by and taken from industrial API manufacturing. The results demonstrate the applicability of the framework for data-driven decision-support on problems such as production planning, line retrofit with project implementation shutdowns and new facility construction under demand and expansion cost uncertainties.