Materials and Methods

Fermenter

Experiments were performed in a pilot scale Plexiglas reactor with an inner diameter (T) of 0.44 m and a working volume of 0.168 m³. The geometrical details of the vessel are displayed in Figure 1. Airflow was introduced through a ring sparger with 36 holes, each diameter of 2 mm, and controlled by a mass flow controller (EL-Flow Select F-203AV, Bronkhorst AG, Netherlands). A 0.065 mol/L Na2SO4 solution was used as model fluid to simulate the non-coalescent behavior present in most fermentation media (Prince & Blanch, 1990).

Experimental methods

The flow regimes were determined by means of the conductivity method described by Bombac et al. (1997). In order to detect the gas-filled cavities behind the impeller blades, and thus, the respective flow regime, a self-made 2-pole conductivity probe with small tip size was applied (Figure 1). The probe was mounted on an in height-adjustable holder, which was moved to the respective impeller level for measurements (Witz, Treffer, Hardiman, & Khinast, 2016). The cavity size and type behind each blade were identified by a drop in the voltage pulse response of the probe. For this, the raw signal was further analyzed by discrete Fourier transformation using PicoScope software (Version 6.9.14.16, Pico Technology Ltd., United Kingdom).
In a first set of experiments, the flow regimes were recorded covering aeration rates between 2 and 18 m3/h and gas superficial velocities (\(u_{g})\) between 0.004 and 0.033 m/s and impeller speeds from 100 to 440 min-1. In a second set of experiments, each impellers transition lines from VC to LC and from LC to RC were identified. For this, the impeller speed was set to a constant value and the gas flow rate was increased in small steps of 0.1 m3/h until a change of the flow regime was recognized. This procedure was repeated at other impeller speeds to be able to draw the transition line. For further analysis of the flow regimes, the transition lines were plotted as a flow map (Warmoeskerken & Smith, 1985), which employs in the dimensionless Froude (Fr) and Flow number (Fl) and allows a coherent classification of the flow regimes.
The gas hold-up and power measurements were conducted in separate experiments with the conductivity measurement setup removed to avoid an influence on the measurements. The gas hold-up and power input were determined for vessel filling levels covering the impeller levels 1, 1 and 2, 1 to 3 and 1 to 4 (Figure 1). The calculation of the difference between values of the reactor filled with \(n\) levels and \(n-1\)levels yielded the value of level \(n\).
The gas hold-up was therefore determined by calculation of the difference of the measured total filling level for aerated conditions for \(n\) impeller levels, \(H_{t,(n)}\), and for the lower \(n-1\)levels, \(H_{t,(n-1)}\), as described by Fujasova, Linek, and Moucha (2007) for \(n=1,\ 2,\ 3,\ 4\) and the relation