About this Abstract |
| Meeting |
2027 TMS Annual Meeting & Exhibition
|
| Symposium
|
Verification, Calibration, and Validation Approaches for Mechanical Modeling of Metals
|
| Presentation Title |
Probabilistic Surrogate Modeling of Thermal Creep in 316H Stainless Steel Under Microstructural Uncertainty |
| Author(s) |
Connor Hopgood, Aaron Tallman |
| On-Site Speaker (Planned) |
Connor Hopgood |
| Abstract Scope |
Data-driven surrogate models (SMs) are being adopted in computational materials science to accelerate materials development across ICME length scales. Though they reduce the computational cost of the constitutive models they represent, SMs have been shown to lose accuracy when the polycrystal mechanistic response reaches a saddle point, after which approximated internal state variables can incorrectly drive the deformation mechanisms. In this work, we propose a probabilistic framework for thermal creep of a steel alloy considered for structural use in the nuclear energy industry. A database of creep evolutions is obtained by sampling uncertain microstructural parameters of 316H stainless steel using the viscoplastic self-consistent method. The resultant SM propagates uncertainty in microstructural inputs through competing deformation mechanisms to quantify the probability distribution of the predicted creep response and identify conditions associated with surrogate imprecision near saddle points. Subsequently, the model is validated in regions where deterministic predictions were not conservative. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Computational Materials Science & Engineering, ICME, Modeling and Simulation |