Knowledge of the radionuclide inventory in spent nuclear fuel is important for back-end operations such as fuel transport and storage, but it is also relevant for the postclosure safety case for a deep geological repository. Extensive depletion calculations using neutron transport solvers can be time consuming and resource intensive in the case of characterization of a large number of fuel assemblies. Issues of computational demand are further amplified when the inventory of only a single pin from the assembly is desired.

A new approach to speeding up the computational time without significant loss of accuracy is proposed in this work, consisting of simplification of the modeled geometry by means of stochastic optimization. The development of this novel methodology, the Acropolis methodology, is described in detail in this paper. Additionally, extensive benchmark and validation exercises were carried out to present and discuss the advantages and limitations of the proposed method.