Sensitivity analysis of nuclear fuel cycle transitions

nuclear fuel cycle;nuclear fuel cycle simulator;sensitivity analysis;transition scenario;Cyclus;DYMOND;Dakota;time series forecasting;automated deployment;d3ploy

Chee, Gwendolyn J ; Stubbins, James ; Huff ; Kathryn D.

关键词: nuclear fuel cycle;    nuclear fuel cycle simulator;    sensitivity analysis;    transition scenario;    Cyclus;    DYMOND;    Dakota;    time series forecasting;    automated deployment;    d3ploy;   

Others  :  https://www.ideals.illinois.edu/bitstream/handle/2142/106272/CHEE-THESIS-2019.pdf?sequence=1&isAllowed=y

美国|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

The present United States nuclear fuel cycle faces challenges that hinder the expansion of nuclear energy technology. The U.S. Department of Energy identified four classes of nuclear fuel cycle options the U.S could transition to, which would overcome these challenges and make nuclear energy technology more desirable. The transitions have been modeled by various nuclear fuel cycle simulators. However, most fuel cycle simulators require the user to define a deployment scheme for all supporting facilities to avoid any supply chain gaps, which becomes tedious for complex transition scenarios. This thesis developed a capability in Cyclus, a nuclear fuel cycle simulator, to automatically deploy fuel cycle facilities to meet user-defined power demand. This new capability successfully deployed fuel cycle facilities in a transition scenario from the current light water reactor fleet to a closed fuel cycle with continuous recycling of transuranics in fast and thermal reactors. In reality, these transition scenarios inevitably diverge from the modeled scenario. This work coupled the nuclear fuel cycle simulator tools, Cyclus and DYMOND, with Dakota, a sensitivity analysis toolkit. This work conducted one-at-a-time, synergistic, and global sensitivity analysis with Cyclus-Dakota and DYMOND-Dakota, to understand the interdependence of input parameters on the transition performance from the current light water reactor fleet to a closed fuel cycle in which transuranics are recycled to fuel mixed oxide fuel thermal reactors and sodium fast-cooled reactors. The global sensitivity analysis concluded that the transition year input parameter was the most influential to the final depleted uranium and total idle reactor capacity performance metrics, and the fleet share ratio and cooling time input parameters were the most influential to the final high level waste amount in the simulation. The one-at-a-time sensitivity analysis showed that varying transition year from 80 to 84 years increased the final depleted uranium amount by 1.13% and reduced the total idle reactor capacity by 10.36%. The one-at-a-time sensitivity analysis also showed that varying fleet share ratio (mixed oxide fuel light water reactor: sodium fast-cooled reactor) from 0:100 to 20:80 reduced the final high level waste amount by 2%, and varying the used fuel cooling time from 0 to 8 years reduced the final high level waste amount by 4%. Therefore, an optimized transition scenario that minimizes final high level waste amount,final depleted uranium amount, and total idle capacity must have a fleet share ratio of 20:80, used fuel cooling time of 8 years, and a transition year at 83 years. This work compared Cyclus-Dakota's and DYMOND-Dakota's sensitivity analysis capabilities and concluded that automated deployment of supporting fuel cycle facilities is crucial for conducting sensitivity analyses with nuclear fuel cycle simulators, to ensure that the simulation adapts to the new parameters by minimizing idle reactor capacity. The results demonstrated that time is saved if a comprehensive sensitivity analysis of a nuclear fuel cycle transition scenario begins with a global sensitivity analysis study to gain a general overview of the influential input variables for the performance metrics. Then, based on the global sensitivity analysis results, a reduced number of one-at-a-time and synergistic sensitivity analyses are conducted to determine quantitative trends and impacts of influential input variables on the performance metrics.