Forensic Measurements for Nuclear Archaeology – A New Approach

Lukas Rademacher - Nuclear Verification and Disarmament Group, RWTH Aachen University
Malte Göttsche - Nuclear Verification and Disarmament Group RWTH Aachen University

The availability of effective and widely accepted verification tools is an essential prerequisite for any lasting and successful effort towards nuclear disarmament. One such verification toolbox is nuclear archaeology - it aims to reconstruct the production and removal history of weapons-usable fissile materials. A central method of nuclear archaeology is the deduction of a shut-down reactor’s lifetime plutonium production using samples taken from within its core. Specific isotopic ratios are measured to assess neutron fluence and thus estimate plutonium production. The method is well established, and since its original inception in the ’90s has seen both practical trial as well as a continuous effort in the community to widen its possible applicability. We will present a new approach aiming to strengthen the potential of the method by analyzing a larger set of measured isotopic ratios. This allows for the reconstruction of operational histories of the considered reactor in more detail, therefore providing more information to crosscheck declarations. The analysis required for this is however much more complex, so we developed a suitable procedure using mathematical and computational methods. As a first step, a computer model of the reactor is created to allow simulation of its neutronic behavior as well as the changes in isotopic compositions of structural reactor elements suitable for the required measurements. We then employ a sensitivity analysis on data obtained from simulations with varied operational histories to identify isotopic ratios containing the information we seek. To keep the computational resources required for the following analysis to a feasible level we introduce Gaussian Process Regression to create a surrogate model for the simulation output of the relevant isotopic ratios. Lastly, the mathematical tool of Bayesian inference is employed to numerically solve the inverse problem and reconstruct operational key parameters from a set of measured isotopic ratios. A feasibility study for this new analysis has been conducted and will be presented, showing that it is indeed possible to reconstruct additional information, such as the reactor's total operational time or power level.