(270b) Modeling and Control of a Rotating Disk Bioreactor

Bioreactors are commonly used to produce biological products of interest, such as cellulose. This work focuses on the development of a rotating disk bioreactor designed to produce not only cellulose, but have the capability to embed particles into the cellulosic matrix. Using a micro-organism, Acetobacter xylinum, cellulose is produced with an acidic byproduct. The modeling and control of the rotating disk bioreactor is a challenging problem due to a high level of coupling and nonlinearities in the system.

A rigorous model of the rotating disk bioreactor is presented, accounting for a dual chamber design. With one chamber for reaction and the other for mixing and measurement, the resulting model is a high order system of n > 10. The presence of distinct time scales allows for singular perturbation theory to be used to group subsystems by time scale and reduce the model order. From this reduced order model, a model predictive control strategy was implemented to simulate the rigorous model, and the results were validated with experimental data. The control objective is to embed particles in a growing cellulose gel with axial and radial specificity, while concurrently maintaining pH and glucose concentration in the reaction chamber at predetermined optimal levels. In order to accomplish this control, the manipulated inputs used are feed rate of concentrated glucose, flow cell dilution rate, caustic feed rate and particle feed rate. Since desired measured outputs such as cellulose thickness and biomass are not readily available in vivo, a Kalman filter is used for state estimation and is integrated into the model predictive control strategy. This allows for unmeasured states to be estimated and used in the control calculations, and results in the ability to embed particles in growing cellulose gel with the desired axial and radial specificity.