Modelling the Influence of Bubble Dynamics on Motion, Mass Transfer and Chemical Reaction in LES-Euler/Lagrange Computations

The Euler/Lagrange approach is an attractive and descriptive method for numerically computing large-scale dispersed multi-phase flows, such as reactive bubbly flows, where however the dispersed phase elements are treated as point-masses. This approach was extended in the present study in order to account for finite size effects, specifically shape and trajectory oscillations as well as the resulting dynamic mass transfer, which are essential in bubble column flows. The flow field was computed by the Large Eddy Simulation (LES) concept with full two-way coupling in momentum and the modelled sub-grid-scale (SGS) turbulence, respecting also bubble-induced turbulence (BIT). Bubble motion was calculated including all relevant forces (i.e. drag, lift, wall force, added mass, fluid inertia, gravity/buoyancy and Basset force), which were extended considering the modelled instantaneous bubble eccentricity and also incorporating bubble transport by the SGS turbulence. Mass transfer was modelled also accounting for bubble dynamic behaviour (i.e. shape oscillations). For validating the model extensions thorough numerical computations were conducted for a number of experimental test cases with only CO2 absorption as well as chemical reactions considering single bubble rise and also bubble swarms in laboratory bubble columns. It is demonstrated that for point-particle approaches the modelling of bubble dynamics in motion and mass transfer is essential for accurate predictions. Only with this extension it is possible to obtain correct bubble lateral dispersion (i.e. bubble fluctuating velocities) and a remarkably higher mass transfer provoked by larger surface area of deformed bubbles. Thereby, the bubble size distribution variation along the bubble column in a reactive system can be predicted with a very good agreement compared to measurements. The transient evolution of species concentration in the column occurred much faster considering the bubble dynamics model resulting in a much better agreement with the measured pH variation.