Multiple impeller reactors are widely used due to their advanced gas utilization and an increased volumetric mass transfer coefficient. However, with the application of Rushton impellers, gas dispersion efficiency varies between the bottom and the upper impeller levels. The present study analyzes the individual flow regime, power input, and gas hold-up in each compartment of a reactor equipped with four Rushton impellers. The results indicate that the pre-dispersion of the air introduced by the bottom impeller (up to 80%) plays a key role in a better gas retention efficiency of the upper impellers (>300%) and leads to a shift of the cavity and flooding lines in the flow map (Fr-vs Fl-Number) of the upper impellers. A novel analysis of the bubble flow in the dispersed state via a two-phase LES-based CFD model reveals that a more homogenous distribution of air bubbles in the upper compartments leads to high compartment gas hold-up values, but fewer bubbles in the vicinity of the impellers. The measured and simulated data of this study indicate that the upper impellers' efficiency mostly depends on the flow regime of and the pre-dispersion by the bottom impeller rather than on the upper impellers' flow regimes. These results contribute to the understanding of essential mixing processes and scaling of aerated bioreactors.

The power input and gas-liquid mass transfer rank among the most important industrial fermentation process parameters. The present study analyzes the power input and gas hold-up as a function of the flow regime, impeller diameter, and rheological properties in a pilot scale reactor (160 L) equipped with four Rushton impellers. This leads to four dimensionless numbers for predicting measurements in pilot and industrial bioreactors (110 and 170 m3) with a standard deviation of 7 % to 29 %. This is unparalleled for the underlying aerated and non-Newtonian fermentation broths. Several existing correlation equations are discussed to be dissatisfying (up to 130 % deviation), and might be sufficiently valid only within scale or for small scaling factors. The introduced approach predicts adequately accurate over three orders of magnitude. Based on these encouraging results, we identified the Galilei number and the power concept as the central elements in combination with the consequent dimensional analysis for an efficient scaling betweeen pilot and industrial scale.