The characterization of radioactive waste from the nuclear industry is essential to ensure their sustainable management, in compliance with French regulations. CEA is developing methods using ionizing radiation to verify the contents of waste packages. Active photon interrogation (API) is a non-destructive technique well suited to large-volume compacted waste, where other techniques fail. It uses an interrogator beam of high-energy photons, capable of inducing fission in nuclear material, in order to estimate its quantity.
The current challenge is to rapidly localize nuclear materials within the package, so that focused spectrometry techniques can then be applied to identify fission products. This is an inverse problem, involving the determination of a radiation source (induced by photofission) from a limited set of measurements. The aim is to develop a Bayesian formalism for dealing with this problem in realistic cases, taking into account the uncertainties related to nuclear data, to the interrogating beam flux, to measurements statistics and to the waste package characteristics (density, composition). Cases of increasing complexity will be studied: homogeneous medium, then heterogeneity, introduction of the 3D density map of the medium obtained by X-ray tomography and the presence of several sources. Experimental validation will be based on the experimental resources of the Laboratoire de Mesures Nucléaires (LMN) at CEA Cadarache, where the thesis will be carried out, including the CINPHONIE irradiation cell where a 15 MeV electron LINAC is installed, nuclear material samples and mock-up waste packages.
This work opens up career prospects particularly in research centers and R&D departments in industry.
A master internship is proposed by the team in addition to the thesis.