Automated algorithm to determine kLa considering system delay

Abstract

Background: Quantification of the volumetric oxygen transfer coefficient (k_L a) is essential to characterize and optimize the oxygen transfer capacity of bioreactors. First order methodologies are commonly used to estimate k_L a; however, when the delay of the system cannot be neglected, second-order methodologies are required for accurate estimations. Second-order methods are time consuming and hard to reproduce. In this study, we describe an automated algorithm to estimate k_L a in conventional bioreactors.Results: The simple, four step algorithm developed considers 1) data smoothing, 2) selection of a high oxygen variation zone, 3) selection of the time period where the instantaneous k_L a is constant and 4) k_L a estimation. The algorithm was coded in MATLAB and four adjustable parameters were fixed heuristically using eight response curves with different hydrodynamic conditions (varying viscosity, agitation and aeration). Compared to manual processing, 80 validation experiments showed that the proposed automatic algorithm yields much more reproducible results in a fraction of the manual processing time.Conclusion: The algorithm is fast and yields, without human intervention, reliable k_L a estimations under different hydrodynamic conditions; hence, it is useful for designing high throughput k_L a assessment systems to optimize oxygen delivery in bioreactors.