Bubble column reactors, where gas is injected at the bottom of an initially stagnant liquid, are widely used in chemical engineering. These reactors often operate in the heterogeneous regime, characterized by a strongly polydisperse distribution of bubble sizes, high gas concentrations (15 to 40% in volume) and strong velocity fluctuations (up to 50% times the mean). Owing to such flow complexity, no reliable scaling rules are available for reactor designers. This problem dramatically hampers optimization and process control. In particular, extrapolation between laboratory-scale and industrial-sized columns (with diameters ranging from 5 to 10 m, and heights between 20 and 40 m) is not controlled, and therefore the simulations lead to realistic results only via an ad-hoc adjustment depending on the size of the column.
In this seminar, the hydrodynamics of bubble columns is revisited with a focus on the increase of the apparent relative velocity between phases observed in the heterogeneous regime. We report experiments in an air-water bubble column of diameter 0.4m using a new Doppler probe that gives access to bubble velocities conditioned by the void fraction. In the first part of the seminar, we will derive and test a new scaling for the liquid velocity. Using our data and results from the literature, we will show that such scaling seems to hold for a wide range of operating conditions and bubble columns with different sizes.
Later, we focus on the characterization of meso-scale structures corresponding to high (clusters), low (voids) and intermediate gas concentrations relative to the mean gas hold-up. These structures are unambiguously identified using Voronoï tessellations built from the phase indicator function. The resulting conditional bubble velocity data demonstrate that meso-scale structures drive the transport of both phases in the heterogeneous regime. The unconditional relative velocity is recovered from conditional relative velocities weighted by the probability of presence of bubbles in a given structure. These results provide a physical origin for swarm factors introduced in simulations. They also offer a way to connect the relative velocity enhancement with meso-scale structures characteristics, and to advance our understanding of scaling rules.