In this work, a DDPM-CFD model is developed in ANSYS® Fluent for the simulation of the indirectly heated, bubbling calciner of the 300kWth dual fluidized bed pilot plant located at Technische Universität Darmstadt. The calciner is heated by 72 heat pipes that carry the heat from an external combustor. Regarding the heat transfer, both convection and radiation are considered in the model. Regarding the modelling of the drag forces, flow heterogeneity aspects are considered by applying the Energy Minimization Multi-Scale (EMMS) scheme. However, the application of DDPM in such dense, bubbling flows considered here proved to be challenging, demanding several advancements and customizations. To this end, this study proposes mainly three advancements; i) The inter-particle forces are modelled using custom user defined functions incorporating both normal and tangential components. In particular, KTGF-based correlations are applied at dilute regions, while at dense regions the solid pressure is modelled according to Harris and Crighton, and the shear and bulk viscosities are modelled using correlations based on the plastic theory. ii) It is shown that, in order to correctly predict the overall pressure drop, the Lagrangian particle momentum equation should be reformulated according to Model A formulation to be consistent with the solved gas-phase momentum equation. iii) In order to capture the correct heat flux levels, the heat flux on the heat pipe heat exchanger walls is modelled in the Eulerian reference frame scaling the temperature gradient on the wall to take into account the thin thermal boundary layer.Τhe DDPM results are compared against those of an already validated Eulerian TFM model, in terms of calculated flow patterns, volume fractions, pressure profiles and heat fluxes. In addition, both models are assessed for their computational cost. The developed DDPM model predicts practically the same overall pressure drop with the TFM model. However, it overpredicts the bed length by 12% when using the default grid. This reduces to 6% when using a finer grid comprising double computational cells. As for the heat fluxes and the calcination reaction rate, both models predict similar levels and their differences are attributed to the differences in hydrodynamics.
Read full abstract