Investigation of nuclear material accountancy produced by small modular reactors

小型模块化反应堆(SMR)具有特殊的燃料组成(如钚、浓缩铀、后处理乏燃料)、新的物理燃料形式(六方晶格、熔融盐、三元粒子等)。以及堆芯中不同的中子能量(先进反应堆概念(ARC-100)和一些熔盐反应堆的快谱)。这些差异的结果是，预计会有不寻常的核材料核算，因此SMR产生的乏燃料预计与加拿大重铀(CANDU)反应堆的乏燃料有很大不同。在过去的40年里，蒙特利尔理工大学开发了用于确定性核反应堆物理模拟的计算机代码DRAGON5(格子代码)和DONJON5(全堆模拟代码)。这项研究计划包括扩展和应用DRAGON5和DONJON5到特定于SMR的新的反应堆物理应用。预计将做出具体贡献：(1)收集与选定的SMR概念(钠快堆、高温快堆和熔盐反应堆)中使用的几何、材料和操作数据相关的信息；(2)将DRAGON5中的共振自屏蔽方法应用于特定的SMR燃料(熔盐、三元粒子等)；(3)将DRAGON5和DONJON5中的贝特曼求解器应用于SMR燃料；(4)将SMR特性纳入类(高级模拟场景核心库)情景模拟器中。CLASS是一个基于神经网络的动态燃料循环模拟代码，可以使用DRAGON5/DONJON5反应堆模型来产生核数据和物理量。这项研究的综合好处将使SMR部署的优化能够最大限度地减少辐射风险，减少扩散问题，评估后处理能力，并更好地了解事故的结果。

Small modular reactors (SMR) feature unusual fuel compositions (e.g. plutonium, enriched uranium, reprocessed spent fuel), new physical fuel forms (hexagonal lattices, molten salts, TRISO particles, etc.) and different neutron energies in the core (fast spectra in Advanced Reactor Concept (ARC-100) and some molten salt reactors). A consequence of these differences is that unusual nuclear material accountancy is expected so that resulting spent fuel from SMR is expected to be very different when compared with spent fuel from Canada Deuterium Uranium (CANDU) reactors.Over the last 40 years, Polytechnique Montréal has developed the computer codes DRAGON5 (lattice code) and DONJON5 (full-core simulation code) for deterministic nuclear reactor physics simulations.This research proposal consists in extending and applying DRAGON5 and DONJON5 to new reactor physics applications specific to SMR. Specific contributions are expected:(1) Gather information related to geometries, materials and operational data used in selected SMR concepts (sodium fast reactor, high temperature fast reactor and molten salt reactor);(2) Adapt the resonance self-shielding methodology in DRAGON5 to the specific SMR fuel (molten salt, TRISO particles, etc.);(3) Apply the Bateman solver in DRAGON5 and DONJON5 to SMR fuel;(4) Include SMR characteristics in the CLASS (Core Library for Advanced Simulation Scenarios) scenario simulator. CLASS is a dynamic fuel cycle simulation code based on neural networks that can use DRAGON5/DONJON5 reactor models to produce nuclear data and physical quantities. CLASS allows the analysis of complex nuclear power plants and of their constituting components.The combined benefits of this research will enable the optimization of the SMR deployment in such a way as to minimize radiological risks, to reduce proliferation issues, to evaluate reprocessing capabilities and to better understand the outcome of an accident.