Implementation of Adaptive Optimal Operation for a Semi-Batch Reaction System

The optimization of fed-batch culture of hybridoma cells is accomplished on a mathematical model using dynamic programming. Optimal feed trajectories are found using a seventh order model for a single feed stream containing both glucose and glutamine and for two separate feed streams of glucose and glutamine. Compared to a constant feed rate, optimal trajectories can improve the final MAb concentration by 11 % for the single feed case and by 20% for the multifeed case. Higher MAb concentrations can be expected for fed-batch optimization with feed enriched in nutrients.

Optimization of fed-batch culture of hybridoma cells using dynamic programming : single and multi feed cases

The dynamic optimization (optimal control) of batch and semicontinuous bioprocesses, like fed-batch fermentation and bioproduct batch separation, has received considerable attention in recent years. These constrained dynamic optimization problems are difficult to solve due to their highly nonlinear character and also due to the fact that they usually are multimodal; therefore most proposed methods may fail to find the global optima. Besides, these methods are not always easy to implement and use, mainly because a laborious preparation and/or transformation of the original problem is needed. Here we present a simple yet powerful method that requires no initial preparation stage and that is useful to solve several complex optimal control problems without convergence difficulties. Two fed-batch bioreactor case studies, taken from the recent literature, and a third case study considering a complex bioproduct dehydration problem are successfully and efficiently solved with our new approach, and its advantages are discussed.

Stochastic Dynamic Optimization of Batch and Semicontinuous Bioprocesses

Optimization of fed-batch bioreactors using genetic algorithm: multiple control variables

There is a need for efficient modeling strategies that quickly lead to reliable models. In the serial gray-box modeling strategy the inaccurately known terms in the macroscopic balance are modeled with a black-box model structure such as a neural network. This way one efficiently obtains an accurate model with good extrapolation properties without the need to develop a rigorous white-box model in which much more knowledge would be involved and without the need to do many identification experiments. Different types of extrapolation are specified and analyzed and they are related to the different parts of the serial gray-box model structure. The strategy is demonstrated for modeling pH effects on the enzymatic conversion of penicillin G using real-life data. The resulting serial gray-box model is compared with a model from a more knowledge-driven white-box strategy and with a model from a more data-driven black-box strategy. The serial gray-box modeling strategy is especially advantageous at the medium level of process operation, which is mainly concerned with the calculation of optimal condition for (bio)chemical processes.

H.C. Lim

Papers

Optimal operation of fed-batch bioreactors with two control variables