Physics-Informed Machine Learning and Uncertainty Quantification for Mechanics of Heterogeneous Materials

A model based on the Physics-Informed Neural Networks (PINN) for solving elastic deformation of heterogeneous solids and associated Uncertainty Quantification (UQ) is presented. For the present study, the PINN Modulus framework developed by Nvidia is utilized, wherein we implement a module for mecha...

Full description

Saved in:
Bibliographic Details
Published in:Integrating materials and manufacturing innovation 2022-12, Vol.11 (4), p.607-627
Main Authors: Bharadwaja, B. V. S. S., Nabian, Mohammad Amin, Sharma, Bharatkumar, Choudhry, Sanjay, Alankar, Alankar
Format: Article
Language:English
Subjects:
Boundary conditions
Characterization and Evaluation of Materials
Chemistry and Materials Science
Elastic deformation
Fiber composites
Finite element method
Heterogeneity
Machine learning
Material properties
Materials Science
Mechanics (physics)
Metallic Materials
Modulus of elasticity
Nanotechnology
Neural networks
Structural Materials
Surfaces and Interfaces
Technical Article
Thin Films
Uncertainty
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:A model based on the Physics-Informed Neural Networks (PINN) for solving elastic deformation of heterogeneous solids and associated Uncertainty Quantification (UQ) is presented. For the present study, the PINN Modulus framework developed by Nvidia is utilized, wherein we implement a module for mechanics of heterogeneous solids. We use PINN to approximate momentum balance by assuming isotropic linear elastic constitutive behavior against a loss function. Along with governing equations, the associated initial/boundary conditions also softly participate in the loss function. Solids, where the heterogeneity manifests as voids and fibers in a matrix, are analyzed, and the results are validated against solutions obtained from a commercial Finite Element (FE) analysis package. The present study also reveals that PINN can capture the stress jumps precisely at the material interfaces. Additionally, the present study explores the advantages associated with the surrogate features in PINN via the variation in geometry and material properties. The presented UQ studies suggest that the mean and standard deviation of the PINN solution are in good agreement with Monte Carlo FE results. The effective Young’s modulus predicted by PINN for single representative void and single fiber composites compare very well against the ones predicted by FE.
ISSN:2193-9764
2193-9772
DOI:10.1007/s40192-022-00283-2