S13: Fed-batch Fermentation Optimization Using Three-Level Definitive Screening Design

Optimization is a critical step in the development of a commercial process. Traditionally, process optimization was achieved through trial and error or by one factor at a time approach. Statistical design of experiments provides a systematic, efficient and economical solution for multivariate optimization. As an initial step of process optimization, resolution III or IV screening designs are used to identify important or critical parameters. Subsequently, a higher resolution response surface design is used to study their effect on the process output, and to predict optimal set points and control ranges. Recently, a novel experimental design approach, called Definitive Screening Design (DSD), was proposed by Jones and Nachtscheim (2011) which is more efficient and economical as compared to the two-step (screening followed by response surface design) approach. Definitive Screening Design allows important factor screening as well as quadratic effect and two-way interaction estimations within a single set of experiments. A 6 factor, 13 run DSD was used to optimize a fed-batch fermentation process, which resulted in doubling the target protein yield. The findings of this study demonstrate the successful application of DSD for process optimization using fewer experimental runs compared to the two-step approach.