Improving prediction of preferential concentration in particle-laden turbulence using the neural-network interpolation

A neural-network interpolation (NNI) is proposed to improve the prediction of preferential concentration in particle-laden turbulence. The NNI uses the particle position and velocity on neighboring grid points to estimate the fluid velocity at the particle position. To evaluate the NNI, we simulate a two-dimensional homogeneous isotropic turbulence subjected to high-wavenumber forcing. The NNI recovers the effect of small-scale motion on particle distribution from the low-resolution field, adding high-wavenumber energy to the turbulence field. Consequently, the NNI improves the prediction accuracy of the preferential concentration on coarse grids.