About this Abstract |
| Meeting |
2026 TMS Annual Meeting & Exhibition
|
| Symposium
|
Verification, Calibration, and Validation Approaches in Modeling the Mechanical Performance of Metallic Materials
|
| Presentation Title |
Sequential Bayesian Parameter Identification for Ductile Damage Models Using Multimodal Experimental Data |
| Author(s) |
Seyed Mohammad Ali Seyed Mahmoud, Surya R Kalidindi |
| On-Site Speaker (Planned) |
Seyed Mohammad Ali Seyed Mahmoud |
| Abstract Scope |
Parameter identification for ductile damage models presents significant challenges due to complex, heterogeneous deformation phenomena and indirect parameter inference requirements. This study presents a Bayesian framework for calibrating Gurson-Tvergaard-Needleman (GTN) damage model parameters using multimodal experimental data combining load-displacement measurements with Digital Image Correlation (DIC) strain fields. A sequential Bayesian updating approach handles computational complexity of high-dimensional multimodal likelihood functions while incorporating experimental uncertainties. Dimensionality reduction techniques extract essential spatial features from DIC data, enabling efficient parameter identification while preserving critical information. The methodology is demonstrated through tensile testing of AA6111 specimens, calibrating four key GTN parameters ($\varepsilon_N$, $f_N$, $f_c$, and $f_f$). Results show the multimodal approach provides improved parameter identification compared to single data stream methods, with quantified posterior distributions capturing parameter correlations and uncertainties. The framework demonstrates potential for calibration of complex constitutive models using readily available experimental techniques, providing a pathway for reliable damage model parameter identification. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Computational Materials Science & Engineering, |