Constitutive Models for the Prediction of the Hot Deformation Behavior of the 10%Cr Steel Alloy

The aim of this paper is to establish a reliable model that provides the best fit to the specific behavior of the flow stresses of the 10%Cr steel alloy at the time of hot deformation. Modified Johnson-Cook and strain-compensated Arrhenius-type (phenomenological models), in addition to two Artificia...

Full description

Saved in:
Bibliographic Details
Published in:Materials 2019-09, Vol.12 (18), p.2873
Main Authors: Shokry, Abdallah, Gowid, Samer, Kharmanda, Ghias, Mahdi, Elsadig
Format: Article
Language:English
Subjects:
Algorithms
Aluminum alloys
Artificial intelligence
Artificial neural networks
Chromium steels
Constitutive models
Correlation coefficients
Datasets
Deformation
High temperature
Magnesium alloys
Mathematical models
Metal forming
Model accuracy
Neural networks
Neurons
Root-mean-square errors
Software
Steel alloys
Strain
Strain hardening
Yield strength
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:The aim of this paper is to establish a reliable model that provides the best fit to the specific behavior of the flow stresses of the 10%Cr steel alloy at the time of hot deformation. Modified Johnson-Cook and strain-compensated Arrhenius-type (phenomenological models), in addition to two Artificial Neural Network (ANN) models were established with the view toward investigating their stress prediction performances. The ANN models were trained using Scaled Conjugate Gradient (SCG) and Levenberg-Marquardt (LM) algorithms. The prediction accuracy of the established models was evaluated using the following well-known statistical parameters: (a) correlation coefficient (R), (b) Average Absolute Relative Error (AARE), (c) Root Mean Squared Error (RMSE), and Relative Error (RE). The results showed that both of the modified Johnson-Cook and strain-compensated Arrhenius models could not competently predict the flow behavior. On the contrary, the results indicated that the two proposed ANN models precisely predicted the flow stress values and that the LM-trained ANN provided a superior performance over the SCG-trained model, as it yielded an RMSE of as low as 0.441 MPa.
ISSN:1996-1944
1996-1944
DOI:10.3390/ma12182873