A novel approach to partial defect testing of spent nuclear fuel for safeguards applications


Pazsit Imre, (Chalmers University (Sweden) ), imre@chalmers.se

SCK•CEN Mentor

Borella Alessandro, aborella@sckcen.be, +32 (0)14 33 28 44

Expert group

Nuclear Science and Technology Studies

SCK•CEN Co-mentor

Rossa Riccardo , rrossa@sckcen.be , +32 (0)14 33 80 35

Short project description

The characterisation of spent nuclear fuel (SNF) assemblies via non-destructive assay (NDA) is very important for the implementation of safeguards, in addition to the safety assessment associated to its handling and disposal.

From the safeguards point of view, SNF is a material of particular interest due to its plutonium content. There are about 270,000 tons of SNF worldwide and the annual arising of used fuel is about 12,000 tons. With an estimated total amount of about 2,700 tons of Pu, it is important to verify that no diversion of SNF has taken place with a very good degree of accuracy.

Current methods for the verification of the integrity of the SNF assemblies rely on the detection of the Cherenkov light when the assemblies are located in spent fuel pool; this method is limited to wet storage and the interpretation of the results is not always straightforward. A recently developed method relies on passive gamma emission tomography; it has the potential of an improved accuracy but the system is expensive, bulky and requires the movement of the SNF from the storage position.

This PhD focusses on the development of a method to detect missing pins in a SNF assembly which is both accurate and does not require the movement of fuel. In addition, due to the reduced detector size, the use for different fuel assembly geometries (e.g. PWR, BWR, VVER) is envisaged. 

The proposed method foresees to perform neutron measurements with movable detectors positioned either among the fuel pins or right next to the fuel assembly. If the spatial dependence of the neutron flux is measured in several radial positions in the assembly, a comparison with the expected neutron flux distribution contains information on possibly missing pins and their positions. For this purpose, it is foreseen to use thin fibre prototype detectors available at Chalmers University.

A first stage involves the improvement of the detectors both in terms of reducing the size while maintaining an acceptable efficiency. In addition, the detectors should be position- (angularly) - sensitive, such that they can be used for the measurement of both the neutron current and the flux gradient. The angularly resolved neutron field is much more effective for identifying the position of localised perturbations, such as a missing pin. Numerical simulations based on Monte Carlo methods will support the optimization of the detector.

The possibility to identify missing pins is based on the deviation between the measured radial space dependence of the neutron flux density in the actual fuel assembly from the one expected from the intact assembly. Since measured data will be available only for the actual fuel assembly under investigation, the expected radial neutron flux and the neutron current in the intact, declared assembly can only be performed by numerical simulations.

Identifying a missing fuel pin or fuel pins is a so-called “inverse task”, i.e. going from the distribution of both the scalar neutron flux and the flux gradient to the actual pin distribution that originated those observables. Since even the direct task can only be solved numerically (either with deterministic numerical methods, or with Monte-Carlo), no self-obvious way of solving the inverse problem exists.

To develop an identification algorithm, we propose the use of non-parametric inversion methods, in particular artificial neural networks (ANNs) for this purpose. A relatively large number of full assembly calculations need to be performed to give a large enough training set for the training of the ANN.

The minimum diploma level of the candidate needs to be

Master of sciences , Master of sciences in engineering

The candidate needs to have a background in

Informatics , Mathematics , Physics
Before applying, please consult the guidelines for application for PhD.