Computational particles, parcels, or super-particles? What are these things?*

The Eulerian-Lagrangian (EL) paradigm has become a leading strategy in the study of particle-laden flows. This method has the advantage of explicitly accounting for the non-continuum dispersed phase while simultaneously accounting for the continuum physics of the carrier fluid. However, practical flows may harbor an enormous number of particles. Therefore, the computational cost associated with simulating each particle can become prohibitively expensive. One strategy to reduce this cost in EL simulations is to simply reduce the number of particles tracked in the system. Each particle tracked in the simulation represents the properties of several particles related to the original system. This strategy, often dubbed the method of ‘computational particles’ is not a new idea. However, there seems to be few systematic studies examining the robustness of this strategy. We will study computational particles in homogeneous turbulence to understand how the ratio of computational to physical particles can be varied along with other parameters such as the Stokes number, volume fraction, and mass loading ratio to ensure that reduction in the number of particles simulated does not result in a significant change in reported statistics of the particle and fluid phases.