(8g) Non-Disruptive Approaches to Estimate Volumetric Mass Transfer Coefficient (kLa) in Real-Time to Obtain Oxygen Uptake Rates (OUR) for Feed Control (Invited Speaker)

Robust and efficient methods to determine the volumetric mass transfer coefficient (kLa) in real-time, that uses only common industrial sensors, are desired. kLa can change in a complex, nonlinear manner during cultures, as it is dependent many parameters, including stir speed, impeller shape and dimension, bioreactor shape and dimensions, media chemical characteristics, cell secretions, and temperature. Estimating real-time kLa without filtering or latency effects allows for the calculation of oxygen uptake rates (OUR) that can be used to control glucose feeding. Currently, the industry standard for feed control is set offline, where the protocol may be adjusted by an operator in response unexpected online or offline measurements, but rarely are online automation used to control feeding to optimal rates. For Escherichia coli, balancing the carbon flux of glycolysis and the tricarboxylic acid (TCA) cycle is critical to minimize acetate accumulation (a waste product) and maximize final cell densities. In this work, an online, state estimator was developed that uses standard industrial sensors to predict the oxygen transfer rate (OTR) in bioreactors, where OTR is directly related to OUR. OUR is an indicator of the metabolic state of the culture and the carbon flux balance. Recursive least squares was used to match a model of OTR to the estimated inlet oxygen concentration, measured dissolved oxygen, and measured off-gas concentrations. The kLa estimator was used to control glucose feeding for E. coli cultures without operator interaction, to culture temperature and induction strength changes. For these E. coli cultures, high-cell density conditions were reached (100 g dry cell weight per L), out competing exponential feeding. Also, initial work has been conducted with Chinese hamster ovary (CHO) cell cultures. For CHO cell cultures, process corrections, such as antiform additions and base additions for pH adjustment, caused disturbances in the off-gas measurements. Work is ongoing to adapt the non-disruptive kLa estimator to CHO cell cultures to match the success of the E. coli cultures.