Optimization methods for prediction of advanced constitutive models


Loading...

Author / Producer

Date

2020-12

Publication Type

Doctoral Thesis

ETH Bibliography

yes

Citations

Altmetric

Data

Abstract

The prediction of strain distribution in production of sheet metal parts directly influences predictions of hardening and failure, as strain and stress response are linked by the yield model. The prediction of strains in simulation is relied upon by tool design and process engineers in all states of development. However, optical measurement and simulation result do not always align, which questions the state of the art approach to define yield models, especially for aluminium alloy. The reason for this mismatch was thought to be the finite complexity of the models used, but, while studying the effect, this work shows that not model complexity but an improved foundation for fitting is a solution. Three advanced fitting strategies are presented for non-quadratic and free-shape yield formulations. The first is based on macroscopic measurement of cruciform tension specimen as specified in ISO16842 for simultaneous evaluation of strain and stress ratio. The second extracts all necessary information by application of inverse full field optimization using digital image correlation and global force measurement. The third method investigates the possibility of crystal plasticity calculation based on texture measurement. In order to evaluate performance and to be able to recommend one strategy for further use, a large validation study is conducted using Nakajima and cross die experiments. Both non-quadratic and free-shape models were able to outperform the original configuration based on the proposed strategies, with cruciform tension and inverse full-field proven to be most effective based on an additional cost analysis. All three strategies provide valuable information and challenge to reconsider common assumptions about the definition of yielding for aluminium alloy. In consequence, this work presents the basis for further research on optimal choice of fitting strategy, which opens up a new perspective on development of yield models.

Publication status

published

Editor

Contributors

Examiner : Hora, Pavel
Examiner : Yoon, Jeong-Whan
Examiner : Mohr, Dirk

Book title

Journal / series

Volume

Pages / Article No.

Publisher

ETH Zurich

Event

Edition / version

Methods

Software

Geographic location

Date collected

Date created

Subject

constitutive modeling; Optimization methods; Digital image correlation (DIC); FEM modeling; Cruciform

Organisational unit

03685 - Hora, Pavel (emeritus) / Hora, Pavel (emeritus) check_circle

Notes

Funding

Related publications and datasets