Modelling and characterization of an airlift-loop bioreactor.

An airlift-loop reactor is a bioreactor for aerobic biotechnological processes. The special feature of the ALR is the recirculation of the liquid through a downcomer connecting the top and the bottom of the main bubbling section. Due to the high circulation-flow rate, efficient mixing and oxygen transfer is combined with a controlled liquid flow in the absence of mechanical agitators.Liquid velocities and gas hold-ups in an external-loop airlift reactor (ALR) on different scales were modelled on the basis of a simple pressure balance. The model is adapted for non-isobaric conditions and takes into account nonuniform flow profiles and gas hold-up distributions across the duct. The friction coefficient together with the reactor dimensions are input parameters. It has been shown that the friction coefficient can be obtained from simple one-phase flow calculations based on known data of the seperate reactor parts. The model predicts liquid velocities and local gas hold-ups in an ALR to within 10% and can be applied easily to an internal loop reactor.Mixing in the individual sections of the ALR is determined by a newly developed parameter estimation procedure which has proven to be reliable for the estimation of axial dispersion coefficients in the individual sections of the ALR. From the results it can be concluded, that in an ALR the liquid flow behaves like plug-flow with superimposed dispersion except for the topsection for which it is not reasonable to assume plug-flow. The mixing results simplified the modelling of oxygen transfer in the ALR as it appeared not to be necessary to incorporate the dispersion contribution Into the oxygen model.The non-isobaric plug-flow model, presented in this thesis, predicts dynamic and stationary dissolved oxygen concentration (DOC) profiles in large-scale ALRs and has been applied also to estimate the volumetric oxygen transfer coefficient, k ₁ a, in the pertinent ALR. Comparison with the results on the basis of a simple isobaric stirred-tank-reactor model demonstrates, that such a model yields conservative values though for the present situation the underestimation did not exceed a value of 10% relative to the plug-flow model. Therefore, due to its simplicity, it is recommended to use the stirred tank model for a rapid characterization of the overall aeration capacity of laboratory scale and pilot-scale ALRs. Oxygen depletion of the gas phase, even during a fermentation, appeared to be very limited and was fairly well predicted by the plug-flow model. For this reason an ALR is a very suitable reactor for aerobic processes having a high oxygen demand. If necessary, the aeration capacity of the ALR can be enhanced by injection of a small amount of gas at the entrance of the downflow region. This phenomenom is also accurately predicted by the plug-flow model. In the present ALR the aeration capacity of the air-sparger region did not significantly differ from the main aeration process in the upflow region due to its special geometry.The intermediate flow region between the ALR and the bubble-column (BC) flow regime was investigated by gradually closing a butterfly valve at the bottom of the downcomer. When the valve is further shut and thus the friction is enhanced, the liquid velocity will be reduced thereby enlarging the gas hold-up. The maximum value for the gas hold-up is obtained when the ALR is operated as a BC. In the transition flow regime between ALR and BC flow, the liquid velocity was found to be a simple power law function of the gas flow rate. The coefficients of the power law depend on the flow characteristics in the reactor. In the transition flow regime the hydrodynamic calculations based on the plug-flow behaviour of an ALR are only valid up to a certain defined value of the total gas-liquid flow rate. For greater values, the ALR type of flow will change Into a BC type of flow. A simple criterium qualifies the distinction between both flow patterns, determined by the superficial liquid velocity and the liquid circulation velocity.The transition of ALR to BC flow coincides with the decrease of the Bodenstein number which also indicates a less established plug flow. As the dispersion coefficient at a constant gas-flow rate, remained constant for as well the ALR, the BC and the transition flow, the decreased Bodenstein number in the BC-type of flow is mainly attributed to the decreased convective transport as the liquid circulation is impeded. The number of circulations required to achieve complete mixing diminshes when the liquid circulation is impeded and appeared to be proportional to the Bodenstein number.In the transition flow regime, the volumetric oxygen transfer coefficient was estimated by both the stirred-tank model and the plug-flow model. The stirred-tank model yielded reliable results for the entire range of operation while the plug-flow model only appeared to be appropiate for the ALR operation mode. The volumetric oxygen transfer coefficient was found to increase for the BC operation mode and appeared to be a power law function of the ratio of the superficial liquid and gas velocity and the Bodenstein number.Addition of immobilized biocatalysts to the ALR, in our case simulated by neutral buoyant particles with diameters ranging from 2.4-2.7 am, significantly reduces the liquid velocity and the gas hold-up in an ALR. The decrease in liquid velocity is attributed to the decrease in gas hold-up and an increased friction in the ALR. The gas hold-up is reduced mainly because the presence of the particles increases the collision frequency of the air bubbles thereby increasing coalescence due to the diminished flowed area available for the air-water mixture. In comparison to a gas-liquid flow, axial dispersion in the three-phase flow is reduced as the presence of the particles damps the small eddies which are, apart from other mechanisms, responsible for the axial dispersion. Moreover. the increased coalescence also contributes to a decrease in axial dispersion. The presence of the particles negatively influences aeration due to a reduction in the gas-liquid interfacial area as a result of the increased coalescence. The effect of the increase in apparent viscosity in the ALR was not supposed to contribute to the decrease in the aeration process