As the world increasingly decarbonises, there is an increasing pressure on chemical manufacturing to move away from fossil carbon sources. Industrial bioprocesses provide one such alternative for fossil carbon. Bubble columns as a bioreactor type are particularly well suited to such large scale applications. However, models of bubble columns, whether based on
design correlations or computational fluid dynamics, have been shown to break down when
air-water systems are replaced with systems containing actual fermentation broth. The additional broth components can significantly affect interphase mass transfer through e.g. limiting bubble breakup, which can in turn make or break the economics of a bioprocess [1]. Experimental data on the effects of broth components on the physics in bubble columns is essential to develop better models. Gathering such data requires experimental methods capable of penetrating the industrially-relevant but opaque churn-turbulent flows. A promising method for determining the state of a large section of the bubble field in a bubble column is X-ray computed tomography. The TU Delft X-ray tomography setup seeks to achieve this using three source-detector pairs capable of capturing X-ray data at high frame rates. However, in order to be able to apply any experimental technique, it must first be validated and the sources of and magnitude of its various measurement errors must be quantified.
This thesis uses computational fluid dynamics to validate tomographic reconstruction algorithms.
The computational fluid dynamics model was validated using experimental data from
Sanyal et al.[2] Furthermore, this thesis finds ways of improving tomographic reconstructions
through discovering which reconstruction algorithms perform best for different datasets. It was found that for time-resolved bubble fields, a version of SIRT (Simultaneous Iterative Reconstruction Technique) with generalised Tikhonov regularisation using the derivative operator performed best with a NRMSE (Normalised Root Mean Squared Error) of 0.0867 over a baseline value of 0.1123 using the default SIRT method and an F-score of 0.641 for the binary
classification of air and water. For time-averaged reconstructions of the gas holdup, an SIRT
with standard Tikhonov regularisation with an offset to the mean gas holdup was found to
perform best with a NRSME of 0.0137 over a SIRT baseline of 0.0160. Finally, this thesis
shows the improvement to tomographic reconstructions for an upgraded version of the TU
Delft X-ray tomography setup and provides recommendations for future research on this topic.
It was shown that increasing the number of source-detector pairs to five, leads to significant
improvements in the time-resolved bubble field reconstructions, with a new NRMSE of 0.0617
(-28%) and F-score of 0.823 (+28%)

The traditional Eulerian view of biomass in bioprocess modelling results in issues when modelling large-scale bioreactors in which heterogeneous conditions are common. As the field increasingly moves from “scale-up” to “scale-down” philosophy, in which such heterogeneities are included from the start of process design, accurate modelling of the biomass response to these varying conditions is essential. Euler-Lagrange (EL) simulations provide a means of modelling the microbial lifelines of cells traversing heterogenous conditions of a bioreactor. Lapin et al. were the pioneers of EL simulation in their 2004 paper where a metabolic model of glycolysis is coupled to a Lagrangian biomass phase. This BSc thesis focusses on reproducing their model using modern computational fluid dynamics (CFD) techniques. Specifically, by using a dynamic Lattice Boltzmann Method using Large Eddy Simulation model for CFD as opposed to a frozen-flow Finite Volume Reynolds Averaged Navier- Stokes model. The resulting differences in the overall behaviour of the cell metabolism through the lens of glycolytic oscillations are discussed. In addition, possible pitfalls in model validity such as grid dependence, the effects of heterogeneous particle distributions and the effects of particle numbers were explored. The synchronisation and desynchronisation of glycolytic oscillations as observed in Lapin et al. 2004 were able to be reproduced.