Large-scale Bioreactor Phenomena and Characteristics; Effects on E.coli fed batch fermentations on 30m3 scale.

Due to economical and practical limitations the fluiddynamics and mass transport properties are less optimal inlarge-scale bioreactors compared to lab-scale bioreactors. Incombination with the microbial reaction rates this creates amuch more dynamic environment for cells cultivated inlarge-scale bioreactors compared to lab-scale bioreactors.To gain more information about these dynamic environmentsand the responses of the cells to circulation in theseenvironments three Esherichia coli fed batch fermentations,each with a different bioreactor configuration, on a 30 m3scale were compared individually and with a lab-scale controlfermentation. The different bioreactor configurations werebased on the use of two different agitation systems (Rushtonand Scaba) and two different feed points for glucose and NH3respectively.Space dependent glucose concentration distributions werefound in the bioreactors. Due to its high degree ofcompartmentalisation the Rusthon agitation system resulted indecreasing glucose concentrations with increasing distance fromthe feed point. For the Scaba agitation system, which had amuch lower degree of compartmentalisation, the spacedistribution of the glucose concentration seemed to be morerandom. The“general”flow pattern in the bioreactorscauses the cells to circulate in the bioreactor and hence thecells experiences relatively long term oscillations in glucoseconcentrations with a frequency determined by the circulationtime. On top of this it was found that“high concentrationglucose packages”can be transported long distances beforebeing diluted and hence the cells to experiences oscillatingglucose concentration with“peak”values severaltimes higher than the mean concentration with a frequency of10-20 min-1 and a duration of 1-5 s.The cells respond to the high glucose concentrations in the“peaks”by overflow metabolism and furthermore theincreased metabolic activity may consume the available oxygenand create local anaerobic zones where mixed acid fermentationis induced. Furthermore space dependent gradients for therelative mRNA concentration of stress genes in the cells werefound in the bioreactors. Together this shows that the cellsrespond very quickly to the dynamics in the bioreactorenvironment. It is proposed that the production andre-assimilation of the by-products were one of the reasons forthe 12,5 % reduction in the biomass yield per glucose in the 30m3 fermentations compared to the lab-scale control. Anotherpossible reason for the lower yield is suggested to be energyloss due to the induction and relaxation of different stressgenes.Fluid dynamic characterisations of the three 30 m3 differentbioreactor configurations were performed to evaluate theeffects of the changes. Even though the fluid dynamiccharacterisation showed significantly better specific powerinput, oxygen transfer and mixing times for the configurationsbased on the Scaba agitation system only small differencescould be seen in the yield coefficients between any of thethree large-scale fermentations. The main differences observedbetween the fermentations were the accumulation of formate athigh cell density and the“restriction”of theglucose oscillations to the“feed zone”for theRushton agitation system. This was suggested to be due to thelower oxygen transfer and the higher degree ofcompartmentalisation compared to the Scaba agitation systemrespectively. Finally, the limited data available indicatesthat the agitation system and the glucose feed point clearlyaffect the glucose oscillation pattern in the bioreactors.However, it does not seem like the better mixing properties forthe Scaba agitation system leads to any significant improvementwhen compared to the Rushton agitation system.