Computer simulations of molten salts interactions with metallic materials

小型模块化反应堆已被加拿大政府认定为“安全、清洁、负担得起的能源来源……为加拿大和加拿大人服务。”这些反应堆概念中有几个依赖于熔盐热交换回路，而不是传统的水基热交换器。一个关键的问题是熔盐驱动的腐蚀鲜为人知。这对于参与设计、规划和建造加拿大小型模块化反应堆的三个主要工业合作伙伴——加拿大核实验室、安大略发电公司和全球第一电力公司来说至关重要。为了弥合这一知识差距，我们将执行原子对原子的计算机模拟，以了解和预测熔盐如何与金属结构材料相互作用。这项研究的一个具有挑战性的方面是，原子间的相互作用本质上是量子的。基于量子力学的模拟方法是准确的，但往往有一个令人望而却步的计算成本。在这里，我们将使用最先进的机器学习算法来构建原子间相互作用模型，其精度可与基于量子力学的方法相媲美，但计算成本可处理。该研究项目将包括在工业合作伙伴的现场进行四个工业实习，帮助培训下一代加拿大核工业科学家。

Small modular reactors have been identified by the Canadian government as a "source of safe, clean, affordable energy [...] for Canada and Canadians." Several of these reactor concepts rely on molten salt heat exchange loops, rather than traditional water-based heat exchangers. A key issue is that molten-salt driven corrosion is much less understood. This is of prime importance for the three main industry partners-the Canadian Nuclear Laboratories, Ontario Power Generation, and Global First Power-which are involved in the design, planning and construction of Canada's small modular reactors. In order to bridge this knowledge gap, we will perform atom-by-atom computer simulations to understand and predict how molten salts interact with metallic structural materials. A challenging aspect of this research is that interatomic interactions are intrinsically quantum in nature. Simulation methods based on quantum mechanics are accurate, but tend to have a prohibitive computational cost. Here, we will use state-of-the-art machine learning algorithms to construct interatomic interaction models with a level of accuracy comparable to that of quantum-mechanics-based methods, but at a tractable computational cost. The research project will involve four industrial internships on the site of the industry partners, helping train the next generation of Canadian nuclear industry scientists.