A hybrid model framework for the optimization of preparative chromatographic processes

Area of Science:

• Chemical Engineering
• Biotechnology
• Analytical Chemistry

Background:

• Preparative chromatography is crucial for purifying biomolecules.
• Optimization of these processes is computationally intensive.
• Existing models may not efficiently handle complex systems.

Purpose of the Study:

• To develop an efficient optimization framework for preparative chromatography.
• To reduce computational time for process simulation and design.
• To enable rapid exploration of various design scenarios.

Main Methods:

• Experimental determination of physical model parameters (general rate model, steric mass action isotherm).
• Development of neural-network-based empirical models using physical model simulation data.
• Optimization using sequential quadratic programming to maximize production rate times yield.

Main Results:

• Hybrid models significantly reduce computational time for simulation and optimization.
• The framework allows for multivariable optimization.
• Rapid exploration of different design scenarios is enabled.

Conclusions:

• Hybrid empirical models offer an efficient approach for complex preparative chromatography.
• This method enhances process optimization and design flexibility.
• The framework is applicable to binary and tertiary model protein systems.