Introduction

Pilot- and large-scale fermentation processes are mostly carried out in bioreactors equipped with multiple impellers. They provide better gas utilization with higher gas phase residence times, an increased gas hold-up, and thus, a higher volumetric mass transfer coefficient (Gogate, Beenackers, & Pandit, 2000). Despite the variety of new impeller models, the most common impeller for microbial fermentation and for educational purposes is still the Rushton turbine. The bottom impeller has been shown to behave differently from the upper impellers regarding flow regime and the resulting oxygen mass transfer coefficient and power input. For instance, several studies showed that the gas hold-up in case of a multiple impeller reactor is higher in the upper impeller compartments than in the bottom impeller compartment (Gogate et al., 2000; Linek, Moucha, & Sinkule, 1996; Nocentini, Magelli, Pasquali, & Fajner, 1988). Gogate et al. (2000) assumed that gas hold-up in the second impeller compartment is about 30 ‒ 40 % higher than in the bottom compartment and Linek et al. (1996) determined an increase of about 15 % on average for the upper compartments, paired with a higher total oxygen mass transfer coefficient of about 40 % in a four-level reactor.
At the same time, numerous studies showed that the bottom impeller exhibits a more pronounced power drop, when aerated than the upper impellers (Hari-Prajitno et al., 1998; Linek et al., 1996; Middleton & Smith, 2004; Warmoeskerken & Smith, 1988). Linek et al. (1996) measured an about 50 % higher power draw by the upper impellers. Nienow and Lilly (1979) assumed that the reason for this behavior is that the number of air bubbles passing the upper impellers must be significantly lower than those at the bottom impeller.
When aerated, Rushton impellers form cavities behind the impeller blades. Extensive studies (Bombac, Zun, Filipic, & Zumer, 1997; Bruijn, Vantriet, & Smith, 1974; Nienow, Warmoeskerken, Smith, & Konno, 1985) analyzed the different forms and transitions of the cavities in relation to impeller speed and aeration rate. In the so-called vortex cavity (VC) regime, which appears at low aeration rates and high impeller speeds, vortices behind the impeller blades disperse the air. In the loaded state at higher aeration rates, the gas accumulates in the low-pressure zone behind each impeller and forms large cavities (LC). With a further increase in aeration rate, the large cavities grow until reaching the next impeller blade resulting in the ragged cavity regime (RC), or also called “flooding”. Typically, flooding occurs earlier at lower impeller speeds. The transitions between these flow regimes have been studied comprehensively for various operating conditions and reactor geometries for reactors with one impeller (Bombac et al., 1997; Lee & Dudukovic, 2014; Lu & Ju, 1989; Nienow, Wisdom, & Middleton, 1977; Warmoeskerken & Smith, 1985). Only few studies investigated the flow regimes in multiple impeller reactors (Abrardi, Rovero, Baldi, Sicardi, & Conti, 1990; Bombac & Zun, 2006; Smith, Warmoeskerken, & Zeef, 1987; Warmoeskerken & Smith, 1988). These studies examined only a narrow operational range and the flow regimes impact has been studied only with regard to power input.
An increasing number of publications analyze the gas-liquid flow in bioreactors by means of computational fluid dynamics (CFD), focusing for example on flow patterns of the gas and liquid phase (Bakker & Oshinowo, 2004; Guan, Li, Yang, & Liu, 2019; Khopkar, Rammohan, Ranade, & Dudukovic, 2005; Khopkar & Tanguy, 2008; Scargiali, D’Orazio, Grisafi, & Brucato, 2007; Wang et al., 2014). Khopkar and Tanguy (2008) investigated the complex pattern in a dual impeller configuration and simulated gas hold-up and the flow pattern of a VC and LC regime. Furthermore, they were able to show a decreasing liquid pumping efficiency of the bottom impeller. However, none of these studies investigated the mechanisms leading to differences between the impeller levels, which were described in the above-mentioned contributions.
The existing data on the flow regimes in multiple impeller reactors and their impact on the gas dispersion behavior is incomplete. Therefore, in this study, we analyze the power input and gas hold-up of each impeller stage and the corresponding flow regime in a four-level reactor and over a wide, industrially relevant operational range. The second objective of this paper is to study mechanisms leading to the differences between the reactor compartments. By means of a two-phase CFD simulation, it is for the first time possible to analyze the bubble flow in each compartment of a pilot-scale (0.15 m3) bioreactor applying high gas hold-up and turbulent conditions, to determine the effective (local) aeration rate and local gas density at each impeller.