Managing trade-offs in protein manufacturing: how much to waste?

T.G. Martagan (Corresponding author), A. Krishnamurthy, P. Leland

Research output: Contribution to journalArticleAcademicpeer-review

1 Citation (Scopus)
53 Downloads (Pure)

Abstract

We consider the challenges and trade-offs involved in the manufacturing of engineered proteins. Manufacturing these proteins involves high risk of financial losses due to the purity and yield trade-offs, uncertainty in the process outcomes, and high operating costs. In this setting, the biomanufacturer must determine how much protein to manufacture in the upstream fermentation operations, and then how much of it to waste in each subsequent purification operation because of the purity–yield trade-offs. We develop a Markov decision model to optimize three layers of interdependent decisions in protein manufacturing: the optimal amount of protein to be produced in upstream operations, the optimal choice of chromatography technique to be used in downstream operations, and the optimal choice of pooling windows during chromatography. The proposed stochastic model dynamically optimizes these three layers of interdependent decisions to maximize the expected profit. The structural analysis derives functional relationships between the purity–yield trade-offs and operating costs, and characterizes the optimal operating policies. The optimal policy also suggests when the biomanufacturer is better off failing early and cutting losses. We use a state aggregation scheme to reduce the computational efforts, and quantify the savings obtained from the use of the optimization model in industry practice at Aldevron.

Original languageEnglish
Pages (from-to)330-345
Number of pages16
JournalManufacturing & Service Operations Management
Volume22
Issue number2
DOIs
Publication statusPublished - Mar 2020

    Fingerprint

Bibliographical note

065-1445

Keywords

  • Protein manufacturing
  • Quality requirement
  • Random yield
  • Stochastic optimization
  • quality requirement
  • protein manufacturing
  • random yield
  • stochastic optimization

Cite this