SIMULATION OF POLYDISPERSE GAS-LIQUID SYSTEMS WITH QBMM: MODEL VALIDATION WITH THREE DIFFERENT TEST CASES

Different industrial scale dispersed gas-liquid systems can be successfully described by considering that gas bubbles are polydisperse, namely the gas phase is characterized by a distribution of bubbles with different velocity, size and composition. Phase coupling issues can be properly overcome by considering the evolution in space and time of such bubble distribution, which is dictated by the so-called Generalized Population Balance Equation (GPBE) (Marchisio and Fox, 2013), with the inclusion of specific mesoscale models for taking into account momentum, heat and mass exchange with the liquid phase, as well as bubble coalescence and breakage. The choice of these mesoscale models is of crucial importance for the prediction of the relevant properties of a gas-liquid system. A computationally efficient approach for solving the GPBE is the quadrature-based moments methods, where the evolution of the entire bubble population is recovered by tracking some specific moments of the distribution and a quadrature approximation is used to solve the “closure problem” typical of moment-based methods. Among these methods, the Conditional Quadrature Method of Moments (CQMOM), is particularly interesting. As pointed out in our previous works on this topic (Buffo et al., 2013b; Buffo and Marchisio, 2014), the development of a very general and fully predictive methodology, useful for simulating different gas-liquid equipments, is the final aim of this work. However, only with reliable mesoscale models for the different interfacial forces, this goal can be achieved. In this work, the focus will be on the relationship for the drag force coefficient, with particular attention to the effect of microscale turbulence and bubble swarm. The proposed methodology is here adopted for the simulation of very different gas-liquid systems, mainly rectangular and circular bubble columns, as well as aerated stirred tank reactors (Laakkonen et al., 2006; Kulkarni et al., 2007; Diaz et al., 2008) and simulation results are eventually compared with the available experimental data.