Erbas, Aarron; (2021) Development and software implementation of modelling tools for rapid fermentation process development using a parallel mini-bioreactor system. Masters thesis (M.Phil), UCL (University College London).

Preview

Text Erbas_10119616_thesis.pdf Download (7MB) | Preview

Abstract

In order to establish a generic framework for the rapid development and optimisation of scalable fermentation processes, a novel methodology for simplifying model building was explored. This approach integrates small-scale fermentations with model-based experimental design (DoE) and predictive control strategies. In this study, four 1.4 litre vessels were characterised for power input, volumetric oxygen transfer coefficient (KLa) and mixing, to assess its potential for replicating cell culture rapidly. Engineering characterisation results showed excellent propeller operation over a range of 400-1200 rpm and up to the maximum motor output and under various air flow rates in fluid densities up to 4.21 Cp/mPa s (1.211 g/cm3 ). Limits were reached using glycerol (99%) at fluid viscosities of 500Cp/mPa s (1.253g/cm3 ) at 800 rpm and no air flow, hence experiencing the most resistance. This was the most taxing condition in terms of energy input into the system. Furthermore, we determined the efficient gas dispersion which is considered important for oxygen bubble dispersion in viscous fluids. The potential gas dispersion could be calculated as a function of both impeller speed, airflow rate, and the fluid viscosity. The calculations provided a working impeller speed of >263 rpm for >0.5 vvm air flow rate as preliminary parameters in our advanced modelling section. The key outcome of the KLa study was that the results showed suitable potential for mass transfer for high cell density fermentations, for each of the parallel stirred tank bioreactors. To assess the usability of the parallel bioreactors be used for bioprocess rapid development purposes Escherichia coli W3110 was characterised in the 1L WV vessels. So overall the experiments included testing the performance of the vessels engineering parameters and also the biological fermentations confirming that the system was suitable for parallel operation with high reproducibility. For model building, especially suited for the 4-reactor set up the parallel bioreactors a fractional factorial design was used, in which models could be rapidly built and implemented for further research. The screening and model optimisation helped to reduce the development time by using the parallel equipment. Batches of four reactors could be completed in parallel in which comparable experimental results were obtained rapidly for new fermentation models. Optical density measurements provided a quick off-line analysis of the growth curve of microbial populations, as compared to cell plate counts or dry weights that require more time. For the model development and the establishment of our integrated software modelling tool, a modified logistic model was developed to predict microbial growth kinetics. First-order kinetic models, logistic, and Gompertz models were used and comparatively analysed to assess the model fit to test batch data. The logistic model was favourable for mapping and simulating the later phases of bacterial growth, while the well-established exponential growth model predicted the early lag phase in our stoichiometric growth simulation software tool better. The initialisation of the previous fermentation model allowed us to build a statistical model, which was based on the engineering characteristics for optimisation of biomass. Therefore, batch nutrient supply with the aid of stoichiometric models could be tested and modelled. DoE model data was improved with metabolic flux analysis to develop an advanced feeding strategy by testing various metabolic pathways and the nutrients used in experimentation. Bacterial growth predictions and media optimisation were tested for maximising microbial biomass yields. We then modelled the dissolved oxygen concentration and substrate utilisation. The techniques and principles of dynamic flux balance analysis, mechanistic modelling, and stoichiometric mass balancing were used. The aim was to create and validate our integrated software based on advanced modelling for the parallel bioreactor systems and tested through application for E. coli fermentations. Optimising microbial biomass was the main target in this project, with the data collected from fermentation being the strongest comparator and validator. A new software for the integration of DoE and Dynamic flux balance analysis (DFBA) techniques with the intention of creating a working fermentation platform for the Multifors equipment via simulation and fermentation optimisation was the novel outcome of this research. The tool could provide functions for speeding up development time and control of parallel bioreactors.

Download activity - last month

Export as XMLCSVJSON

Download activity - last 12 months

Export as XMLCSVJSON

Downloads by country - last 12 months

Export as CSVXMLJSON

Archive Staff Only

View Item