| About this Abstract |
| Meeting |
2024 TMS Annual Meeting & Exhibition
|
| Symposium
|
Materials Informatics to Accelerate Nuclear Materials Investigation
|
| Presentation Title |
Utilizing Mechanistic Modeling and Uncertainty Analysis to Support Nuclear Fuel Qualification |
| Author(s) |
Christopher Matthews, Michael WD Cooper, Pieterjan Robbe,, Habib N Najm, David Andersson |
| On-Site Speaker (Planned) |
Christopher Matthews |
| Abstract Scope |
By breaking down material behavior to the fundamental behavior of defects, multi-scale models can be built that are based in physical reality, but functional enough to extend to extreme environments, i.e., high temperature and high dose. Through the implementation of atomisticly derived data in a cluster dynamics code, we are able to predict the diffusivity of fission gas in nuclear fuel, achieving quantitative comparison for commonly used fuels, as well as prediction of behavior in the absence of data for exotic fuel types. These mechanistic models are then extended with machine learning techniques to cascade uncertainty quantification to integral, measurable predictions of defect species behavior. The combination of mechanistic modeling with Bayesian approaches facilities rapid fuel qualification by quantifying model uncertainty. This in turn provides clear guidelines for experimental campaigns or modeling efforts to decrease prediction uncertainty, and ultimately operational margin. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Computational Materials Science & Engineering, Nuclear Materials, Modeling and Simulation |