The project team has consisted of experts from several different research areas: from reactor physics and artificial intelligence, to computational physics and experimental reactor physics. An advisory group of end users has ensured that the research has been carried out in line with the needs of the nuclear power industry and that the benefits of the innovations can be used in the industry.
The team's collaboration has led to the development and testing of a technology that can detect disturbances in nuclear power reactors.
Combines nuclear reactor modelling and artificial intelligence
“By our combined expertise, we have achieved a technology that combines nuclear reactor modelling and artificial intelligence by which you can detect if there is an anomaly in a reactor core. The technology can also detect what kind of disturbance there is and where in the system it is located,” says Christophe Demazière.
The fundamental of the technology is to teach an artificial intelligence algorithm how a nuclear reactor behaves in the presence of different types of disturbances and their positions. These disturbances lead to fluctuations in the neutron flux, the so-called neutron noise, and they are measured by neutron detectors in the reactor. The algorithm needs to be fed with a lot of data of different types of disturbances and corresponding responses from the reactor.
“To build such a database, we have developed advanced modelling tools. The algorithm then compares the measurements from the reactor with simulations from these modelling tools. From all these simulations, the algorithm can thus identify in a given measurement if there is a disturbance, of what type it is and where it is located. A reactor core is around three to four meters in diameter and height. Using a few neutron detectors in the core, we can detect where there is a disturbance with a margin of five to ten centimetres. Previous research has shown that this is something that could be done, but no technology has been developed to do so in such a systematic way and to such an extent as in the Cortex project,” says Christophe Demazière.
Keeps track of disturbances
The technology can, for example, be used during operation to see what is happening in the reactor, the so-called core monitoring. By keeping track of disturbances, you can also better plan for how to handle possible problems when closing a reactor for inspection, maintenance, and fuel reloading.
Further development of the technology will be required before it can be used on an industrial scale. How or in what form the research project will continue remains to be seen.
How has it been then, to coordinate such a large project, with so many participants?
“In the beginning, we spent a lot of time getting to know each other's different research fields, in order to work towards the same goal. We have had close contacts with each othereach other, and everyone has been very motivated in this collaboration. It has been a great job to contribute to the project and see that you do something useful,” says Christophe Demazière.
Facts about the research project:
Cortex (CORTEX) stands for "core monitoring techniques and experimental validation and demonstration." The project aimed to develop innovative methods that can be used to detect and categorize disturbances in commercial nuclear reactors during operation. The method is non-intrusive. Cortex is a research and innovation project (RIA) within the
EU program Euratom in Horizon 2020. Read more about the project on
Cortex's website.
The project has been coordinated by Professor Christophe Demazière and Associate Professor Paolo Vinai, and also involved Dr. Antonios Mylonakis and PhD student Huaiqian Yi, all from the division of Subatomic, High Energy and Plasma Physics at the Department of Physics at Chalmers University of Technology. The researchers have contributed with knowledge in the field of reactor modelling and core monitoring, within which there is a long research tradition at Chalmers where Professor Imre Pázsit’s contributions and influence have been crucial.
