Verification and validation of a DEM-CFD model and multiscale modelling of cohesive fluidization regimes

Abstract

Fluidization of solid particles using gas flow is an important process in chemical and pharmaceutical industries. The dynamics of fluidisation are intricately related to particle scale physics. Fluid-particle interactions dominate gas-solid fluidization behaviour for particles with average size and density greater than 10-4 m and 103 kg/m3, respectively, classified as Geldart B and D particles. Inter-particle forces, such as cohesion, play an increasingly important role in the fluidization dynamics of smaller particles, which are classified as Geldart A and C. In particular, interesting fluidization regimes have been noticed for weakly cohesive Geldart A particles, exhibiting a window of uniform fluidization before the onset of bubbling behaviour. Despite widespread industrial interests, the fundamental understanding of the mechanisms that underlie these fluidization regimes is poor. The present study aims to improve the understanding of fluidization dynamics of Geldart A regimes using numerical simulations. A DEM-CFD model was employed to capture the widely separated spatial and temporal scales associated with fluidization behaviour. The model couples the locally averaged Navier-Stokes equation for fluid with a discrete description of the particles. The methodology and its computer implementation are verified and validated to assess the extent of fluidization physics that it is able to capture. Verification cases check the implementation of the inter-phase momentum transfer term, drag model implementation and pressure-velocity coupling. The test cases are employed in order to cover a wide range of flow conditions. Robust validation tests for complex fluidization phenomena such as bubbling, spouting and bidisperse beds have been conducted to assess the predictive capabilities of the DEM-CFD solver. The simulation results for time and spatially averaged fluidziation behaviour are compared to experimental measurements obtained from the literature, and are shown to have capture fluidization physics qualitatively. Robust features of bubbling fluidization, such as minimum fluidization velocity, frequency of pressure drop fluctuations, segregation rates and solid circulation patterns were captured. Furthermore, the DEM-CFD model is critically assessed in terms of model conceptualization and parameter estimation, including those for drag closures, particle-wall boundary conditions, bed height and particle shape effects. The validation studies establish modelling best-practice guidelines and the level of discrepancy against the analytical solutions or experimental measurements. Having developed the model and established its predictive capability, it is used to probe the hydrodynamics of weakly cohesive particles. Cohesive interactions are captured by employing a pair-wise van derWaals force model. The cohesive strength of the granular bed is quantified by the ratio of the maximum van der Waals force to the particle gravitational force, defined as the granular Bond number. The Bond number of the bed is increased systematically from 0-10 to examine the role of cohesion in the fluidization behaviour of fine powders while keeping the particle size and density constant across all the simulations. The idea was to segregate the hydrodynamics associated with size and density of the particles from the inter-particle interactions. The size and density of the particles are carefully chosen at a scale where inter-particle forces are present but minimal [Seville et al., 2000]. The Geldart A fluidization behaviour is captured for granular beds with Bond numbers ranging from 1 to 3. Many robust features of Geldart A fluidization, such as pressure drop overshoot, delay in the onset of bubbling, macroscopic Umf predictions and uniform bed expansion are captured in the DEM-CFD framework. The expanded bed was characterized according to criteria that the particles are highly immobile in this regime and the expanded porosity is related to inlet velocity by Richardson–Zaki correlations. Sudden jumps in the magnitudes of global granular temperature were found near the regime transitions. This observation was used an indicator of the onset of bubbling and quantification of minimum bubbling velocity (Umb). The window of the expanded bed regime (quantified as Umb - Umf) was shown to be an increasing function of cohesive strength of the bed. Furthermore, the stability of the expanded bed was probed by studying the response of the expanded bed to sudden inertial and voidage shocks. A kinematic wave, generated as a response to the voidage shock, was shown to slow down with increasing cohesion and decreasing hydrodynamic forces. Furthermore, predictions of Umb by DEM-CFD simulations for weakly cohesive beds were compared against empirical correlations by Valverde [2013] with an excellent match. Stress analysis of the expanded bed revealed the presence of tensile stresses. As the inlet velocity is increased beyond the minimum fluidization velocity, a longitudinal shift of these negative stresses is observed until they reach the top of the bed. Negative stresses were seen at the bed surface at the onset of bubbling. The role of cohesion stresses in the formation of expanded bed and suppression of bubbling was highlighted. Finally, the microstructure of the expanded bed was probed at different local micro and mescoscopic length scales. Evidence of clustering, agglomeration and cavities were presented in the expanded bed. Expanded bed expansion was shown to have mesostructural inhomogeneities present, which is contrary to the belief of homogeneous expansion.