Eulerian-Lagrangian Simulations of Fuel Mixing in Fluidized Beds

Licentiate thesis, 2011

Fluidized-bed technology has been commercially applied over several decades. However, there is still a lack of knowledge that can provide detailed understanding of the combustion process in the furnace. Especially, understanding of mixing of fuel particles is crucial in order to be able to optimize the number of fuel feeding ports as well as to lower the excess air. Thus, it is important to develop tools for reliable design and scale up of fluidized-bed boilers, including modeling from first principles by Computational Fluid Dynamics (CFD). In fluidized-bed boilers, there is typically a low mass fraction of large fuel particles in an inert bed of finer solids. The fuel particles are considerably larger than the inert bed material and a conventional Lagrangian Particle Tracking method is not able to correctly handle this difference in size. The method would typically use a single grid in which the size of computational cells is based on the size of the fuel particles. In such a case, it is not possible to accuratly resolve the flow of the carrier phase and the behavior of the inert particles. Therefore, in the present work, a three-grid method, including a fine, a coarse and a moving grid, is proposed. The fine grid is employed to resolve the flow of the carrier phase and to treat the small (inert) particles, whereas the coarse grid is used to calculate the drag forces acting on the fuel particles. Furthermore, moving cells are used in order to correctly calculate the pressure gradient force on the fuel particles. To investigate the performance of the methodology, several numerical cases are simulated. Using a statistical analysis developed in the present work, preferential positions and the dispersion coefficient of the fuel particles under different operating conditions are obtained. The detailed motion of the fuel particles in the form of upward and downward velocity is studied. In addition, two types of boundary conditions are tested; a uniform velocity profile at the distributor and a non-uniform velocity profile obtained by including the air-supply system in the computational domain. The numerical results are compared to available experimental results.