Dynamic optimization of bioprocesses: efficient and robust numerical strategies

Area of Science:

• Bioprocess Engineering
• Chemical Engineering
• Computational Science

Background:

• Non-linear bioprocesses are often described by complex differential-algebraic equations (DAEs).
• Solving dynamic optimization problems for these systems presents significant numerical challenges due to non-linearity, constraints, and discontinuities.
• Existing solution techniques often struggle with the inherent complexity and non-convexity of these problems.

Purpose of the Study:

• To review and address the numerical difficulties in dynamic optimization of non-linear bioprocesses.
• To present novel stochastic and hybrid techniques based on the control vector parameterization (CVP) approach.
• To evaluate the efficiency and robustness of these new methods compared to existing approaches.

Main Methods:

• The study employs the control vector parameterization (CVP) approach, transforming dynamic optimization problems into non-linear programming (NLP) problems.
• A hybrid technique is introduced, combining a global optimization phase with a local deterministic method to handle non-convexity.
• The methods are tested on challenging case studies involving fed-batch bioreactors and other bioprocesses.

Main Results:

• The presented stochastic and hybrid techniques, particularly the two-phase hybrid approach, effectively overcome numerical difficulties.
• The CVP-based methods transform complex DAE systems into solvable NLP problems.
• The two-phase hybrid method demonstrates a superior balance of robustness and computational efficiency.

Conclusions:

• The two-phase hybrid optimization technique provides a robust and efficient solution for the dynamic optimization of non-linear bioprocesses.
• Control vector parameterization is a viable strategy for converting complex bioprocess control problems into non-linear programming tasks.
• This research offers improved computational tools for optimizing bioprocess operations, enhancing efficiency and reliability.