Nuclear Fuel-debris Characterisation via Multimodal Spectroscopy and Analytics (NuFAMSA)

On 11th March 2011, a major Earthquake followed by a 15-metre tsunami caused an unprecedented accident in the Fukushima nuclear power plant. Fukushima Dai-ichi reactors suffered core meltdowns in the following days. Since then, complex activities for the removal of the melted fuel have been planned, where one of the main challenges is the characterisation of nuclear fuel-debris, including its detection, safeguard, retrieval and disposal.

This research brings together a complementary and multidisciplinary expert team of researchers and industrialists from the UK and Japan to explore the use of hyperspectral imaging (HSI) along with other sensor technologies and data fusion for the effective characterisation of nuclear fuel-debris.

Conventional visual techniques have already been used for remote visual assessment of the core debris in the reactors at Fukushima. Some spectroscopy approaches, able to collect data across the electromagnetic spectrum have already been proposed as more powerful techniques to distinguish fuel-containing debris from non-fissile core spall via their respective spectral fingerprints. These include Laser Induced Breakdown Spectroscopy (LIBS), Raman, gamma-ray spectroscopy, and neutron assay. However, these techniques are limited to point-based measurements, a key limitation leading to no spatial information of the scene under analysis being captured. Multiple point-based measurements across a spatial location are possible, but at the cost of being extremely time-consuming, within small areas, and with a poor spatial resolution. In practical terms, these techniques are unable to scan a spatial location, and only spectral data can be collected. To address this limitation, HSI is introduced in this research.

HSI is able to simultaneously capture the spatial and spectral content of a given scene under analysis. Unlike conventional images, which capture three channels (Red, Green and Blue) in the visible range of the spectrum, HSI captures hundreds of channels covering not only the visible but part of the infrared range (normally 400-2500nm), going beyond what the human eye can see. In fact, in a hyperspectral image, the content of each pixel is a vector array containing the spectral response or fingerprint of the material captured by that pixel. Therefore, HSI is able to capture spectral information across a spatial location, exposing the different fingerprints of the materials present in the scene under analysis, and their distribution.

The core idea of this project is to use HSI technology to generate spatial maps in which fuel-containing debris can be distinguished from non-fissile core spall, avoiding re-criticality during debris retrievals. Based on the different spectral fingerprints, it is also expected that HSI can be used to further characterise certain waste types and automatically detect them within an accurate location. Moreover, HSI can be combined with other sensor technologies, where HSI can be used as a pre-screening tool to direct the point-based acquisition of LIBS, gamma-ray spectroscopy neutron assay. We anticipate that the proposed research will lead to new and highly valuable inspection technology which can support nuclear decommissioning in Japan, the UK and around the world.