Credibility of Predictions of Reactor Core Condition
Typically, about 20% of the UK’s electricity is supplied from nuclear power-stations [http://gridwatch.co.uk/] which are operated by EDF. These power-stations are approaching the end of their planned life so that a number of them are usually shutdown for inspection and maintenance [https://www.edfenergy.com/energy/power-station/daily-statuses]. A key factor in their future operation is the condition of the graphite in the reactor which moderates the neutrons and also forms the structure of the core. During inspection periods, the inner surface of the graphite core is examined and measured to support models of the degradation of the graphite during the operation of the reactor [https://www.youtube.com/watch?v=hlaUky_v55I].
The inaccessibility of most of the core means that measurement data is limited to a few locations whereas predictions about the condition of the entire core are required to support decisions about the safe operation of the reactor. A credibility matrix and associated credibility factors have been developed for use in evaluating untestable models in aerospace engineering and computational biology; and, it is proposed to extend and modify these tools to establish trust amongst decision-makers in predictions of the condition of the graphite core in nuclear reactors.
The project will consider strategies to maximise the use and collection of measurements from graphite cores, as well as to improve protocols for communicating information about the quality of a model including its uncertainty. The project will be conducted in a research group involved in similar work in aerospace engineering and biotechnology; and industrial supervision will be provided by EDF and Frazer-Nash Consultancy.
Project Description In this project, it is proposed to consider approaches to establishing credibility for predictions of the spatial distribution of mass loss in the graphite core of nuclear reactors. This will involve: reviewing positions in the philosophy of science and their relevance and applicability to predicting mass loss in the core of nuclear reactors; exploring the applicability and usefulness of the approaches to establishing confidence and credibility for models used in other fields, including aerospace engineering, computational biology and climate change modelling; developing strategies to maximise the use of existing measurement data and for the collection of further data; and improving protocols for communicating information about the quality of a model and its predictions to decision-makers, including data about uncertainty. The latter will build on current research on validation metrics [Dvurecenska et al, Royal Soc. Open Science, 5:180687, 2018] and fields of uncertainty.
If time permits, models of the distortion and cracking of graphite bricks will be considered as well as of mass loss.
It is proposed that the academic supervision will be supplemented by an industrial supervisory team from EDF and Frazer-Nash Consultancy. Meetings of the complete team will be held quarterly to review progress and discuss the future direction of the project as well as to provide expert advice and information to support the project.
To develop and apply concepts for establishing credibility in predictions of the condition of the graphite core in nuclear reactors.
Objectives 1. To review positions in the philosophy of science and their relevance and applicability to predicting mass loss in the graphite core of nuclear reactors;
2. To explore the applicability and usefulness of the approaches to establishing confidence and credibility for models in other fields, including aerospace engineering, computational biology and climate change modelling;
3. To develop strategies to maximise the use of existing measurement data and for the collection of further data from graphite cores;
4. To improve protocols for communicating information about the quality of a model and its predictions to decision-makers, including data about uncertainty.
Training The first period of Student Specific Training [SST1] will be undertaken at the University of Liverpool and will include: a review of the existing scientific literature in the field of model credibility and the validation of computational models; and a validation exercise involving predicting and measuring the shape and deformation of a simple test sample in order to gain practical experience of the issues associated with acquiring and comparing data, quantifying uncertainty, establishing confidence and communicating to decision-makers. Taught modules in computational modelling and uncertainty analysis might be included depending on the background and experience of the student recruited.
In SST2, the student will be seconded to EDF and, or Frazer-Nash in order to gain experience and understanding of both the models and experiments that underpin the current approaches to predicting the condition of the graphite core of reactors. If the opportunity is available, a visit to NNL Central Laboratory to observe the measurements made on the trepanned samples from graphite cores will be arranged. The outcome of this period of training will be a comprehensive understanding the problem and associated issues that will be tackled in the PhD.
Proposed supervisors (lead first): Eann Patterson (University of Liverpool)
Industrial supervisor/partner (if applicable): Bruce Davies (EDF) & Andy Moore/Mark Joyce (Frazer-Nash)
CDT name: GREENCDT