Benchmarking a Novel 0-D Model Against Data from Two-Fluid Model Simulations of a Wet Fluidized Bed

Research output: Chapter in Book/Report/Conference proceedingConference paperpeer-review


Processes relying on wet fluidized beds (FBs) are widely used in various industries such as the petrochemical and the food & pharmaceuticals sector, for several purposes: i) to improve the particle flowability, taste, and appearance; ii) to modify the particle density, or the particle size distribution; and iii) to protect the particles against humidity, light, oxygen etc. Consequently, it is required to consider wetting, drying, particle shaping and size enlargement effects, as well as the homogenization of the product. Fluidized beds allow integrating all of the above purposes into a single process step, mainly due to the high rate of heat and mass transfer in these devices.1 Various phenomena occur in wet fluidized beds that need special consideration: i) deposition of droplets on the particle surface; ii) evaporation of liquid from the particle surface; iii) evaporation of free-flowing droplets; and iv) particle agglomeration (not considered in this study). Since contributions of these exchange rates are difficult to be experimentally investigated, the numerical study of such system can be beneficial. In our present study, a Two-Fluid Model (TFM)-based code (MFiX2) was extended to consider gas-particle-droplet interactions in a small-scale wet fluidized bed. Since the computational cost is a big challenge in simulation of industrial beds on a system level, a novel 0-D model was developed. The accuracy of this simplified model was examined by quantifying the uniformity of the particles’ LoD (Loss on Drying) distribution in the bed as simulated by the TFM. The results of TFM simulation showed that evaporation and deposition phenomena are mainly limited to the spray zone, while drying happens in the dense bed. This leads to an almost uniform distribution of LoD in the bed in case no droplets are washed out of the FB with the fluidization air. To quantify the degree of uniformity of LoD, the standard deviation of LoD was computed in the bed based on TFM data. It was demonstrated that this standard deviation is 2 orders of magnitude smaller than the total change of LoD during the flow time. As a consequence, the bed can be assumed to well-mixed as long as no droplet loss happens. Several sets of simulations were performed considering various spray rates and initial bed temperatures using both the TFM and the 0-D model. The macroscopic bed characteristics from TFM simulation were imported to the 0-D model simulations. It was proven that the key process parameters (i.e., LoD, gas humidity, and gas temperature) could be well predicted by the 0-D model. However, in case droplet loss occurs as a consequence of a too small droplet injection velocity, the 0-D model fails to predict the outcome of our TFM simulations. Specifically, the largest deviation between TFM and 0-D model prediction was observed for bed temperature and particle LoD. Also, deviations between the model predictions for very shallow beds were unveiled by our simulations. This highlights the important effect of the solid flow pattern on the mixing rate in the bed, and hence the ability to assume a well-mixed particle phase in FBs in the context of system level models. 1Mörl L, Heinrich S, Peglow M. Fluidized bed spray granulation. Handbook of powder technology. 2007;11:21-188. 2Syamlal, Madhava, William Rogers, and Thomas J. OBrien. MFIX documentation theory guide. No. DOE/METC--94/1004. USDOE Morgantown Energy Technology Center, WV (United States), 1993
Original languageEnglish
Title of host publicationAIChE Annual Meeting 2018
Publication statusPublished - 29 Oct 2018
Event2018 AIChE Annual Meeting - Pittsburgh, Pittsburgh, United States
Duration: 28 Oct 20182 Nov 2018


Conference2018 AIChE Annual Meeting
Country/TerritoryUnited States


Dive into the research topics of 'Benchmarking a Novel 0-D Model Against Data from Two-Fluid Model Simulations of a Wet Fluidized Bed'. Together they form a unique fingerprint.

Cite this