Date of Award

12-2018

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Electrical and Computer Engineering (Holcomb Dept. of)

Committee Member

Dr. Richard Groff, Committee Chair

Committee Member

Dr. Sarah Harcum

Committee Member

Dr. Rod Harrell

Abstract

For the last several years drugs based on monoclonal antibodies have been manufactured using Chinese Hamster Ovaries (CHO) cells by the bio-pharmaceutical industries to treat cancer and other autoimmune diseases. Several control strategies are used to increase the productivity and efficiency in bio-pharmaceutical manufacturing. Cell growth can be controlled by adjusting the feed rate based on oxygen uptake rate of the cells in the bioreactor. Determining the volumetric mass transfer coefficient and oxygen saturation concentration is vital in correctly estimating oxygen uptake rate. Thus, a robust and efficient method to determine volumetric mass transfer coefficient and oxygen saturation concentration, which uses common industrial sensors, is desired. In this thesis, a new method to determine volumetric mass transfer coefficient is proposed and implemented on simulated and laboratory experiments. Using this method, volumetric mass transfer coefficient can be calculated independently of oxygen saturation concentration. The fitting parameters required to estimate volumetric mass transfer coefficients are estimated using only the estimated oxygen mole ratio of input gas, the measured oxygen mole ratio of the off-gas and the dissolved oxygen concentration in the bioreactor. A modified version of Savitzky-Golay filtering is used to determine the change in oxygen concentration in the bioreactor liquid. Another algorithm is used to reduce the variations between estimated OUR ( OUR ) and OURlinfit signal to estimate the oxygen saturation concentration in the liquid. Finally, both these signals are used to estimate final OUR signal. The performance of these algorithms were validated by simulated experiments and lab experiments. A Simulink model was used to simulate bioreactor experiments and the values obtained after implementing the algorithm on simulated experiment data were compared with known values from the Simulink model to verify algorithm accuracy. High accuracy was obtained in all the simulated experiments even in presence of noise. The variation and noise in estimated OUR was significantly reduced when these algorithms were employed. The algorithm could also be used in cases when there were sudden gas mix changes by estimating OUR using parameters estimated just prior to the gas mix change. The algorithm was applied to laboratory experiments and it showed consistent results over short periods of time. Since the oxygen saturation concentration is important information required to estimate OUR and control the growth rate of cells, these algorithms have the potential of proving useful in implementing robust controller to increase the productivity and efficiency of the monoclonal antibody manufacturing process using CHO cells.

Share

COinS