NE Colloquium - Zineb Aly - "Computation Models in Nuclear Engineering"
Computational models are widely used in engineering to approximate physical phenomena and enhance our understanding of complex physics. Nuclear engineering is a multi-disciplinary area combining multiple scientific fields. The continual improvement of computational tools leads to the use of complex numerical codes in nuclear engineering where the physics models depend on a large number of parameters. These parameters propagate their own uncertainty in the code predictions and impact the output accuracy. Sensitivity and Uncertainty (S/U) analysis provides a method to quantify the input-output relationships of computer codes to enhance our understanding of the code predictions and accuracy.
In this work, we demonstrate a global S/U approach to quantify the impact of uncertainty in the hydrogen migration and redistribution models implemented in the U.S. Department of Energy Office of Nuclear Energy fuel performance code Bison. In this study, we provide a brief description of the physical phenomena studied and the sensitivity analysis methods used. To identify the key parameters related to the hydrogen migration and redistribution model in Bison, we study the impact of the variance of the model parameters on the amount of hydrides formed near the outer surface of the nuclear fuel cladding, where hydrides are more likely to form, under the normal operation conditions of a light water reactor. To quantify the impact of the input variance of the parameters on the output variations, we compute the variance-based indices
(Sobol indices) and the Pearson correlation coefficients. The results of this work show that the activation energy for the terminal solid solubility of hydride precipitation, the hydrogen heat of transport and the activation energy for hydrogen diffusivity are the key parameters. An optimized set of these parameters was then determined as an attempt to increase the accuracy of Bison predictions by decreasing the root mean square error of the predictions versus experimental results, using a basin-hopping optimization framework.
Wednesday, February 27, 2019 at 1:30pm to 2:30pm
Nuclear Engineering Building, 302
1412 Circle Drive Knoxville TN 37996