Multi-physics reactor optimisation coupled with uncertainty quantification

由于涉及到许多不同的尺度和耦合的物理现象，核反应堆系统的建模是复杂系统建模中最具挑战性的任务之一。如果还要考虑到核数据、操作条件、制造公差、年龄变化等方面不可避免的不确定性，任务就会更加复杂。随着代码能力的进步和高性能计算资源的可用性，现在可以进行一次性计算的分析，但是在设计优化过程中多次执行此类分析的计算成本仍然令人望而却步。为了便于设计优化，必须采用计算成本较低的模型。这些模型可以是使用有限数量的高保真度（计算成本很高）模拟或基于简化物理的低保真度模型构建的替代品。该项目将探索采用最先进的替代建模和多保真度分析方法来促进核系统的多物理场分析和设计优化的方法。

The modelling of nuclear reactor systems is one of the most challenging tasks in complex system modelling, due to the many different scales and coupled physical phenomena involved. The task is further complicated if efforts are also made to account for the unavoidable uncertainties in nuclear data, operating conditions, manufacturing tolerances, changes with age etc. With advances in the capabilities of codes and in the availability of high performance computing resources, it is now possible to undertake such analyses as one-off calculations, but the computational cost of performing such analyses many times within a design optimisation process remains prohibitive. To facilitate design optimisation computationally cheaper models must be employed. Such models can be surrogates built using a limited number of high-fidelity (computationally expensive) simulations or lower-fidelity models based on simplified physics. This project will explore ways in which state-of-the-art approaches to surrogate modelling and multifidelity analysis can be employed to facilitate multi-physics analysis and design optimisation of nuclear systems.