Search

Based upon the two fluid model (TFM) theory, a CFD model was implemented to investigate a cold multiphase-fluidized bubbling bed reactor. The key variable used to characterize the fluid dynamic of the experimental system, and compare it to model predictions, was the time-pressure drop induced by the bubble motion across the bed. This time signal was then processed to obtain the power spectral density (PSD) distribution of pressure fluctuations. As an important aspect of this work, the effect of the sampling time scale on the empirical power spectral density (PSD) was investigated. A time scale of 40 s was found to be a good compromise ensuring both simulation performance and numerical validation consistency. The CFD model was first numerically verified by mesh refinement process, after what it was used to investigate the sensitivity with regards to minimum fluidization velocity (as a calibration point for drag law), restitution coefficient, and solid pressure term while assessing his accuracy in matching the empirical PSD. The 2D model provided a fair match with the empirical time-averaged pressure drop, the relating fluctuations amplitude, and the signal’s energy computed as integral of the PSD. A 3D version of the TFM was also used and it improved the match with the empirical PSD in the very first part of the frequency spectrum.

In this work, CFD simulations of an air-water bubbling column were performed and validated with experimental data. The superficial gas velocities used for the experiments were 0.019 and 0.038 m/s and were considered as an homogeneous regime. The former involves simpler physics when compared to a heterogeneous regime where the superficial velocities are higher. In order to simulate the system, a population balance model (PBM) was solved numerically using a discrete method and a closure kernels involving the Luo coalescence model as well as two different breakup models: Luo's and Lehr's. For the multi-phase calculations, an eulerian framework was selected and the interphase momentum transfer included drag, lift, wall lubrication, and turbulent dispersion terms. A sensitivity analysis was performed on a Luo coalescence kernel by changing the coalescence parameter (c0) from 1.1 to 0.1 and results showed that the radial profiles of gas holdup and axial liquid velocity were significantly affected by such parameter. From the simulation results, the main conclusions were: (a) A combination of the Luo coalescence and Luo breakup kernels (Luo-Luo) combined with a decreasing value of c0 improves the gas holdup profiles as compared to empirical values. However, at the lowest value of c0 investigated in this work, the axial liquid velocity deteriorates with regards to experimental data when using a superficial gas velocity of 0.019 m/s. (b) A combination of the Luo coalescence and Lehr breakup models (Luo-Lehr) was shown to improve the gas holdup values with experimental data when compared to the Luo-Luo kernels. However, as c0 decreases, the Luo-Lehr models underestimate the axial liquid velocity profiles with regards to empirical values. (c) A first and second order numerical schemes allowed predicting similar radial profiles of gas holdup and axial liquid velocity. (d) The mesh sensitivity results show that a 3 mm mesh size can be considered as reasonable for simulating experimental data. (e) The inclusion of wall lubrication parameter was found to be significant, although only when using finer meshing. In addition, it allows an improvement of the axial liquid velocity at the core of the bubble column.