A high resolution collision algorithm for anisotropic particle populations
Pischke, Philipp (Corresponding author); Kneer, Reinhold
Aachen : Publikationsserver der RWTH Aachen University (2014)
In turbulent particle laden flows such as liquid sprays, droplet collisions make a signicant contribution to momentum transfer and energy dissipation. By Lagrangian particle tracking with the stochastic parcel method, only a computational subset of the particle population is simulated, known as computational parcels; the prediction of particle collisions in based on a statistical assessment of collision probabilities. Prior to the preparation of this work, various collision algorithms have been investigated with respect to exactness, robustness, and convergence. Within these preliminary studies, two signicant errors were identied, namely voidage errors and gradient errors, which are unresolved in any collision algorithm found today. To address these issues, a hybrid deterministic-stochastic formulation for the collision probability has been derived, which treats parcel collisions in a deterministic manner, i.e. based on their trajectories and distances, while the collisions of the represented particles are predicted in a stochastic manner. The methods derived are not introducing additional numerical parameters, i.e. spatial and temporal resolution are dened by the number of parcels and the numerical time-step only. The formulations have undergone a rigorous analytical and numerical validation, and show advanced accuracy and convergence behavior when compared to other formulations for stochastic collision algorithms. The methodology, convergence tests, and exemplary applications are subject to this presentation.