Hybrid Uncertainty Propagation on Advanced Nuclear Fuel Cycles

Fuel cycle simulators are versatile tools for the modeling of different electronuclear scenarios, allowing for the study of their impact in terms of viability, waste management or costs. Nevertheless, since they are fed by a large number of parameters, it is not trivial how the uncertainties in the input are propagated through the simulations. Moreover, due to the cumulative effect they may have, the scenario outcomes can be drastically changed. Although traditionally uncertainty quantification has been studied by means of local methods (sensitivity coefficients, a first order approximation), nowadays when high level of computing performance is widely available, other more general theories like the Sobol indices methodology gain attraction since they employ no approximations. In this work, both approaches have been coupled for the study of an advanced nuclear scenario with the use of the TR_EVOL code, which is developed and maintained by CIEMAT. The first part of the study verifies that the minimization and posterior stabilization of all the minor actinides produced in an advanced fuel cycle can be achievable with the combination of a mixed fleet of light water reactors and ADS systems (as the objectives of the scenario demand). Results in terms of mass inventories are presented, as well as their uncertainties and the subset of most relevant variables with the highest contribution to it. Finally, from the differences between the sensitivity coefficients and the Sobol indices, it can be concluded that, for this scenario, the selected outputs can be approximated by a linear and separable model.