(106c) Mixing Simulations for the Scaling-up of Succinic Acid Production from Biorefinery Glycerol

The main objective of this work is to investigate the hydrodynamics in a biochemical process at different scales to improve its performance. The scaling-up of a bioprocess from a small scale to a large scale reduces the main  product yield  mostly due to the alteration of the mixing behaviour in different reactor sizes. It is observed that the mixing efficiency is reduced with increasing reactor size. Scaling-up of biochemical systems results in insufficient gas supply, noncontrolled pH and substrate gradients,  due to unsatisfactory mixing (Oosterhuis and Kossen, 1984; Larsson et al., 1996). Furthermore, reaction times are often shorter than mixing times in lab-scale reactors in contrast with  industrial scale reactors where mixing times are much larger (Akiti and Armenante, 2004).     

Recently, Theodoropoulos and co-workers (Vlysidis et al., 2011; Rigaki et al., 2013) have developed a (fed)batch reactor model for the prediction of succinic acid (SA) production from glycerol in order to calculate the substrate, product and byproduct(s) consumption/production rates. The constructed ordinary differential equation (ODE) based model was deemed to accurately predict the concentration transients in small-scale reactors. Nevertheless, it was observed that scaling-up resulted in differences between the model predictions and the experimental observations, mainly due to differences between the small and the large bioreactors’ principal hydrodynamics. For this purpose, this study focuses on the behaviour of macro-mixing in the scaling-up of such a stirred tank bioreactor system. Macromixing accounts for the flow processes manipulating the mean concentration and the residence time needed for the molecules to flow through the reactor (Vicum et al., 2004).      

Computational fluid dynamics (CFD) simulations are performed here, employing COMSOL Multiphysics, to investigate the  mixing importance in such systems and to complement the system parameters’ optimisation as well as the robust system design and control. The proposed CFD model couples the k-ε model used for the simulation of  turbulent kinetic energy (k) and of energy dissipation rate (ε), with convection/diffusion as well as the reaction processes taking place in the fermentation system. The resulting set of partial differential equations is solved simultaneously leading to efficient prediction of the concentration transients in the volume of the agitated fed-batch bioreactor.

References

Akiti O. and P.M. Armenante, AIChE Journal, 50:566-577, 2004.

Larsson G., M. Tornkvist, E.S. Wernersson, C. Tragardh, H, Noorman and S.O. Enfors, Bioprocess and Biosystems Engineering, 14:281-289, 1996.

Oosterhuis N.M.G. and N.W.F. Kossen, Biotechnology and Bioengineering, 26:546-550, 1984.

Rigaki A., C. Webb and C. Theodoropoulos, Chemical Engineering Transactions, 35:1033-1038, 2013.

Vicum L., S. Ottiger, M. Mazzotti, L. Makowski and J. Baldyga, Chemical Engineering Science, 59:1767-1781, 2004.

Vlysidis A., M. Binns, C. Webb and C. Theodoropoulos, Biochemical Engineering Journal, 58-59:1-11, 2011.