Porous object drawer

Model

PIPN

PIPN++

PIPN++ MRG

PI-GANO

PI-GANO++

Spline degree

Object points

Darcy coefficient

Inlet velocity m/s

Inlet angle

Simulating...

An error occurred! Please try again.
Reduce the sharpness of the porous object corners or increase the object size.

About This webapp allows to experiment with the 2D models presented in the paper Modeling diffusion in complex media through Physically Informed Neural Networks.
The goal of the paper was the simulation of variable mixed fluid-porous domains using Physics Informed Neural Networks.
The predictions are computed using the Physics Informed PointNet or the Physics Informed PointNet++.
Both the predictions obtained from the models and the ground truth simulations resulting from OpenFOAM are available, together with the Mean Absolute Error and the Navier-Stokes-Darcy equations residuals, calculated using autograd.
To draw a custom porous object:

Drag the control points to set the shape
Right click to add a new control point
CTRL-Right click to delete a control point

Got it!