Determination of material properties of bulk metallic glass using nanoindentation and artificial neural network

An artificial neural network (ANN) model is combined with finite element (FE) nanoindentation to evaluate free volume model (FVM) parameters for bulk metallic glass (BMG). FVM is numerically implemented with the user material subroutine (UMAT). A material database is generated based on FE analysis,...

Full description

Saved in:
Bibliographic Details
Published in:Intermetallics 2022-05, Vol.144, p.107492, Article 107492
Main Authors: Park, Soowan, Fonseca, João Henrique, Marimuthu, Karuppasamy Pandian, Jeong, Chanyoung, Lee, Sihyung, Lee, Hyungyil
Format: Article
Language:English
Subjects:
Amorphous materials
Artificial neural network
Artificial neural networks
Bulk metallic glass
FEA
Finite element method
Free volume model
Latin hypercube
Material properties
Mathematical models
Metallic glasses
Nanoindentation
Neural networks
Parameters
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:An artificial neural network (ANN) model is combined with finite element (FE) nanoindentation to evaluate free volume model (FVM) parameters for bulk metallic glass (BMG). FVM is numerically implemented with the user material subroutine (UMAT). A material database is generated based on FE analysis, in which indentation parameters are obtained from FVM parameters. An ANN is generated in order to correlate FVM and indentation parameters and trained/tested from the generated database after the application of removal of multicollinearity, sampling, and normalization for computational efficiency. The fully trained ANN inversely evaluates the FVM parameters from the indentation parameters. The ANN approach is experimentally validated by sphero-conical/Berkovich indentation load-depth curves of Zr55Cu30Ag15 and Zr65Cu15Al10Ni10. •ANN model for evaluating material properties of bulk metallic glass.•Practical application of free volume model.•Use of Latin hypercube sampling for high efficiency.•Cross-validation with sphero-conical and Berkovich indentations.
ISSN:0966-9795
1879-0216
DOI:10.1016/j.intermet.2022.107492