Digital Twin Model Development for Through-Life Management of the Divertor in a Fusion Power Plant

2017年，我在约克大学完成了物理学本科学位。后来，我决定与脑损伤等有复杂需求的人合作，发现这是非常有意义的。2018年，我的职业生涯从数据角色开始，先是在一家临床试验测试实验室，然后在一家医疗和温控包装公司。2020年，我在牛津的一家医学成像软件初创公司工作，在那里我帮助建立了深度学习神经网络模型，以描绘CT和MR扫描中的结构。从2022年开始，我一直在核工程行业担任数据科学家，之所以进行这一转变，是因为我从大学开始就一直对聚变能源感兴趣。在我找到了一个既符合我的数据技能又符合我的物理知识的项目后，这让我转向了Fusion CDT。分束器是MCF中的一个关键方面，特别是在使用H模等离子体的球形托卡马克中。偏滤器能够通过排出杂质和散热来处理X点以外的等离子体效应。因此，它将封装托卡马克的一个区域，在其中可以找到关键的诊断信息。然而，目前偏滤器是一个设计、建模和控制非常复杂的部件。因此，有必要开发偏滤器的数字孪生部件，以捕获模型和数据，以提供反应堆更完整的图像。这将有助于了解材料和组件的状态，提醒用户进行预测性维护，并使用等离子体和面向等离子体的材料的模型运行假设场景。混合建模方法的一个关键方面将是对多物理模型和数据驱动模型进行定量验证，并通过物理测试和量化不确定性来确定故障预测的可信度。这些将是未来聚变能源工厂未来设计、制造、实施和终身管理的非常有用的技能。这个博士项目由全球核工程公司Assystem共同资助，Assystem是一家专门从事工程、数字和项目管理的全球核工程公司。ASSystem在核裂变和聚变方面都有知识和经验，例如支持英国原子能机构在库勒姆的聚变实验，以及它作为热核实验堆建筑师工程师的角色。

I completed my undergraduate degree in Physics at the University of York in 2017. Afterwards, I decided I wanted to work with people with complex needs such as brain injuries and found this incredibly rewarding. In 2018, I started my career in data roles, first in a clinical trials testing laboratory and then at a medical and temperature-controlled packaging company. In 2020, I worked at a medical imaging software start-up in Oxford where I helped build deep learning neural network models to delineate structures within CT and MR scans. From 2022, I have worked as a data scientist within a nuclear engineering industry, making this transition because I have always been interested in fusion energy since university. This led me towards the Fusion CDT after I found a project which fit both my data skills and physics knowledge.Divertors are a key aspect within MCF, especially in spherical tokamaks using a H-mode plasma. The divertor is able to handle plasma effects beyond the X-point via exhausting impurities and removing heat. It will therefore encapsulate an area of the tokamak in which key diagnostics can be found. However, currently, the divertor is a very complex component to design, model and control. Therefore, there is a need to develop a digital twin of the divertor to capture models and data to provide a more complete picture of the reactor. This will help understand the state of the materials and components, alert users to predictive maintenance and run what-if scenarios using models of the plasma and plasma-facing materials. A key aspect of the hybrid modelling approach will be quantitative validation of multi-physics and data-driven models and establishing credibility in failure predictions through physical testing and quantifying uncertainty. These will be very useful skills for future design, manufacturing, implementation and through-life management of future fusion energy plants.This PhD project is being co-funded by Assystem, a global nuclear engineering company with specialities in engineering, digital and project management. Assystem has knowledge and experience in both nuclear fission and fusion such as supporting UKAEA's fusion experiments in Culham and in its role as architect engineer for ITER.