|
Optimizing Terminal Time of Fed-batch Bioreactor with Model Predictive Control |
|
Tae Hoon Oh(Kyoto University, Japan) |
| Optimizing the terminal time is a crucial factor in enhancing the fed-batch bioreactor's productivity. To date, this optimization problem has been addressed by Dynamic Programming or Pontryagin's Minimum Principle-based approaches. However, the discretization of the given input trajectories causes significant errors in actual implementation. To address this problem, this paper proposes to use Model Predictive Control (MPC) to optimize the terminal time with discrete time control inputs. The simulation results show that MPC can successfully improve productivity by adjusting the terminal time of fed-batch operation. |
|
|
Model-based Design and Control of Biopharmaceutical Manufacturing Processes |
|
Moo Sun Hong(Seoul National University, Korea), Richard D. Braatz(Massachusetts Institute of Technology, United States) |
| Recent trends in biopharmaceutical manufacturing provide opportunities for process systems engineering to make major advances in biomanufacturing. To address the challenges associated with the recent trends, this article describes how modern system engineering tools are applied to develop advanced biomanufacturing systems. For the upstream process, mathematical models and optimal control methods are constructed for multiple bioreactor configurations. For the downstream process, laboratory unit operation systems are designed and constructed for protein crystallization and continuous viral inactivation based on first-principles models. |
|
|
Tube-Enhanced Multi-Stage Model Predictive Control for Robust Fed-Batch Cultivation of E. Coli |
|
Jong Woo Kim(Incheon National University, Korea) |
| We discuss the application of an adaptive tube enhanced multi-stage model predictive control (MPC) to achieve an robust optimal operation of E. coli fed-batch cultivations with intermittent bolus feeding. 24 parallel experiments were considered in a high-throughput microbioreactor platform at a 10 mL scale. During the cultivation, the dissolved oxygen concentrations have to be always above the specific lower boundary to prevent the cell from being exposed to anaerobic conditions. To maintain such condition under uncertain model parameter value, we use a robust MPC strategy called tube enhanced multi-stage MPC. We present the simulations for bacterial fed-batch cultivation. |
|
|
Batch Bayesian optimization of culture conditions for Yarrowia lipolytica to produce lipid as a biofuel precursor |
|
Kyeongsu Kim(KIST, Korea) |
| A systematic approach to optimize the culture conditions of newly created Y. lipolytica strain is suggested. The batch Bayesian optimization is adopted to simultaneously find plural experimental conditions for each round, which enable a high-throughput experimental design. One round of experimentation tests 20 conditions of glucose, xylose and ammonium concentrations for Y.lipolytica culture. The experimental design is performed by the diversity-guided Bayesian optimization by modifying the acquisition function to be suitable for batch evaluations. |
|