其他摘要:The implementation of a Lagrangian particle transport model in non-homogeneous turbulent flow is presented. The model proposes one-way interaction between the continuous and the discrete phases where the behavior of the continuous one is previously solved with other CFD software. Particle dynamics include four forces terms: buoyancy, inertial, drag and added mass. Also, a Discrete Random Walk (DRW) formulation is added to model the changes on the trajectory of the particles due to turbulence. The temporal integration is carried out using a Runge-Kutta-Felhberg (RKF) integrator, and a novel analytical integrator is used to solve particle-wall collisions. To manipulate the time-step on each particle, a three-layer filter is developed: an user layer, a RKF layer and a physical layer determined by the particle relaxation time. Also, it is proposed an efficient implementation that uses shared memory techniques to parallelize the execution. To validate the developed code, laminar and turbulent academic tests are presented. Finally, experimental test data from a pilot plant are reproduced using the in house code along with a commercial CFD software, showing a good agreement with data and a radical improvement in time computing comparing with the CFD software.