Fuel Performance Code Benchmark for Uncertainty Analysis in Light Water Reactor Modeling

Open Access
Author:
Blyth, Taylor Scott
Graduate Program:
Nuclear Engineering
Degree:
Master of Science
Document Type:
Master Thesis
Date of Defense:
None
Committee Members:
  • Maria Nikolova Avramova, Thesis Advisor
  • Kostadin Nikolov Ivanov, Thesis Advisor
Keywords:
  • Fuel Performance Codes
  • UAM Benchmark
  • Uncertainty Analysis
Abstract:
Fuel performance codes are used in the design and safety analysis of light water reactors. The differences in the physical models and the numerics of these codes along with input, manufacturing, and boundary condition uncertainties can lead to more variations in predicting the target parameter. Because of this, an uncertainty analysis is an important step in code development and testing. Determining the best estimate values with confidence bounds of important fuel quantities are becoming a more essential benchmark of the fuel performance codes. An uncertainty analysis, such as performed in this thesis, targeting the common sources of variation in the fuel performance codes shows the effects of uncertainty in manufacturing tolerances and boundary condition variations on the centerline temperature of the fuel. This is done with an uncertainty analysis code, DAKOTA, driving simulations of randomly sampled variations in input parameters, as defined by the UAM Benchmark, coupled with the fuel performance codes FRAPCON and FRAPTRAN. The input parameters with the strongest influence on the output are also identified. With 100 simulated cases for each test problem, the overall minimum and maximum calculated output values were within 6% of the calculated sample mean of the output parameter. The fuel density variations had the largest impact on the calculated fuel centerline temperature. The results in this study show that the variations of the input parameters is propagated to the calculated target parameters and best estimate values along with confidence bounds can be used to define the expected results with 95% confidence. As a result, a benchmark for fuel performance codes has been designed for these types of cases.