Prediction of fluid flow in porous media by sparse observations and physics-informed PointNet.

We predict steady-state Stokes flow of fluids within porous media at pore scales using sparse point observations and a novel class of physics-informed neural networks, called "physics-informed PointNet" (PIPN). Taking the advantages of PIPN into account, three new features become available compared to physics-informed convolutional neural networks for porous medium applications. First, the input of PIPN is exclusively the pore spaces of porous media (rather than both the pore and grain spaces).

Mechanical trapping of particles in granular media.

Mechanical trapping of fine particles in the pores of granular materials is an essential mechanism in a wide variety of natural and industrial filtration processes. The progress of invading particles is primarily limited by the network of pore throats and connected pathways encountered by the particles during their motion through the porous medium. Trapping of invading particles is limited to a depth defined by the size, shape, and distribution of the invading particles with respect to the size, shape, and distribution of the host porous matrix.