Online optimization of dynamic binding capacity and productivity by model predictive control.

Area of Science:

• Biochemical Engineering
• Chromatographic Process Optimization
• Separation Science

Background:

• Current chromatography processes face a trade-off between dynamic binding capacity, column utilization, and productivity.
• Optimizing the loading step is crucial for process economy, balancing productivity, resin utilization, and buffer consumption.

Purpose of the Study:

• To present an online optimization approach for capture chromatography using a residence time gradient during loading.
• To improve the traditional trade-off between productivity and resin utilization in chromatographic processes.

Main Methods:

• Employed an extended Kalman filter as a soft sensor for real-time product concentration monitoring.
• Utilized a model predictive controller with a pore diffusion model for online optimization.
• Implemented a residence time gradient during the loading step.

Main Results:

• Achieved significant improvements in productivity and resin utilization.
• Demonstrated buffer savings of up to 43% compared to traditional methods.
• Validated the approach with monoclonal antibody and lysozyme purification systems, showing robustness to a 50% variation in feed concentration.

Conclusions:

• The online optimization strategy effectively maximizes productivity and resin utilization.
• This approach offers substantial buffer savings and enhances overall process economy in chromatography.
• The method is robust and applicable to various chromatographic systems, including affinity and ion-exchange chromatography.