Computational analysis of propagation of geometrical uncertainties to tritium production in FNG WCLL experiment

We present a computational sensitivity and uncertainty analysis of the effect of geometrical uncertainties on
the measured reaction rates in the Water Cooled Lithium Lead experimental benchmark, which is a mock-up
of a tritium breeding blanket for the future DEMO fusion facility. We used deterministic transport methods
to determine the sensitivity of various parts of the geometry and Monte Carlo particle transport methods
along with random sampling from geometrical uncertainties, to assess the expected uncertainty and sensitivity profiles, particularly to establish whether tritium self-sufficiency can be achieved within the uncertainty. We have established that most sensitive parts of the geometry are the source position and orientation, as well as the target detector pack position. We have established the uncertainty to be in the range of 8 % to 25 %, increasing with the distance from the source. We have also established the sensitivity to the displacement of the target detector pack to be roughly 5 % mm−1, while the sensitivity to the perturbation of position of other detector packs and source position and orientation is generally below the 1 % mm$^{−1}$.