Microstructure-Informed Probabilistic Model for Hydrodynamic Forces in Particle-Laden Flows

We propose a new model for the hydrodynamic forces exerted on a spherical particle in a randomly distributed monodisperse array of spheres for a Reynolds number range of $ 2 \le Re \le 150 $ and a solid volume fraction range of $ 0.1 \le \phi \le 0.4 $. By constructing probability distribution maps extracted from direct numerical simulations, our model takes advantage of the statistical information of the arrangement of neighbors to correlate the microscale variation of the drag and lift force with the local anisotropy of each particle’s neighborhood. Given the locations of the neighboring particles as input, our results demonstrate that the present probabilistic model is capable of predicting of up to 70\% of the observed force variation. Since precise location of each particle is known in an Eulerian-Lagrangian simulation, our model would be able to estimate subgrid force fluctuations reasonably well, and thereby greatly enhance the fidelity of mesoscale simulations through improved interphase coupling.