Integrating upstream and downstream planning to increase biomanufacturing profitability
Abstract
Problem definition: Biomanufacturing uses biological systems, such as living cells, to produce an increasing number of products, including biopharmaceuticals. Industrial large-scale production occurs in two stages: the upstream fermentation process, where the product is formed in a nonlinear and stochastic manner, and the downstream purification process, where specialized resins are used for purification but exhibit stochastic capacity decay. Despite their operational interdependence, both stages are often managed separately, neglecting the trade-offs between upstream production and downstream purification capacity and their impact on profitability.
Methodology/results: We develop a stochastic optimization model that integrates upstream harvesting, downstream purification, and resin replacement decisions using available process information. The problem is formulated as a discrete-time, infinite-horizon Markov decision process that maximizes long-term profitability. Exploiting the sigmoidal structure of product formation, we analytically characterize the optimal policy and show that there exist control-limit structures with intuitive bounds. We further compare integrated and separate decision-making and identify the economic conditions under which integration creates value. Numerical results for large-scale insulin production show that the integrated policy increases profits by approximately 4% compared to a condition-based but separate approach.
Managerial implications: The optimal policy can be implemented as simple threshold-based rules using standard process measurements. Our results show that profitability requires coordinated decision-making across upstream fermentation and downstream purification, and that focusing solely on upstream productivity or batch yield can lead to suboptimal outcomes. The value of integration is particularly high in low-margin environments, such as biosimilar production, where downstream costs become economically critical.

