Modeling and Online Estimation of kLa and P/V for 250mL and 5L Mammalian Bioreactors

In this study, we quantify the volumetric mass transfer coefficient of oxygen (ka) and specific power input (P/V) in bioreactors. These are essential parameters that define a bioreactor's operating conditions and are used to address bioprocess development challenges, namely, developing scale-up/scale-down models and tech-transfer. These parameters are difficult to measure in real-time.Using two datasets from Culture Biosciences' 250-mL and 5-L mammalian bioreactors, we modeled kLa and P/V based on three variables: agitation, air sparging flow rate, and reactor volume. Three modeling techniques were explored: polynomial regression, a power and gassing kLa correlation, andGaussian Kernel Regression. The non-parametric Gaussian Kernel Regression demonstrated flexibility in capturing complex, non-linear relationships. We utilize Gaussian Kernel Regression models to compare kLa, P/V, and oxygen transfer rates between 250-mL and 5-L bioreactor scales fora Chinese Hamster Ovary cell (CHO) process. These kLa and P/V models allow existing clients to gain insight to scale-up related parameters, and assist new clients in transferring their processes ontoCulture’s platform.