We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

physics.flu-dyn

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Physics > Fluid Dynamics

Title: Physics-Informed Neural Network Method for Forward and Backward Advection-Dispersion Equations

Abstract: We propose a discretization-free approach based on the physics-informed neural network (PINN) method for solving coupled advection-dispersion and Darcy flow equations with space-dependent hydraulic conductivity. In this approach, the hydraulic conductivity, hydraulic head, and concentration fields are approximated with deep neural networks (DNNs). We assume that the conductivity field is given by its values on a grid, and we use these values to train the conductivity DNN. The head and concentration DNNs are trained by minimizing the residuals of the flow equation and ADE and using the initial and boundary conditions as additional constraints. The PINN method is applied to one- and two-dimensional forward advection-dispersion equations (ADEs), where its performance for various P\'{e}clet numbers ($Pe$) is compared with the analytical and numerical solutions. We find that the PINN method is accurate with errors of less than 1% and outperforms some conventional discretization-based methods for $Pe$ larger than 100. Next, we demonstrate that the PINN method remains accurate for the backward ADEs, with the relative errors in most cases staying under 5% compared to the reference concentration field. Finally, we show that when available, the concentration measurements can be easily incorporated in the PINN method and significantly improve (by more than 50% in the considered cases) the accuracy of the PINN solution of the backward ADE.
Comments: 31 pages, 15 figures
Subjects: Fluid Dynamics (physics.flu-dyn); Computational Physics (physics.comp-ph)
DOI: 10.1029/2020WR029479
Cite as: arXiv:2012.11658 [physics.flu-dyn]
  (or arXiv:2012.11658v1 [physics.flu-dyn] for this version)

Submission history

From: Qizhi He [view email]
[v1] Mon, 21 Dec 2020 19:48:17 GMT (21234kb,D)

Link back to: arXiv, form interface, contact.