球床式高温气冷堆运行过程中，石墨包覆的球形燃料元件由于摩擦等作用会产生石墨粉尘，石墨粉尘在冷却剂氦气的携带下在一回路内运动并沉积在设备表面上。当反应堆变工况运行或发生破口事故时，沉积在壁面上的石墨粉尘会在气流作用下发生起尘现象而脱离壁面，由于石墨粉尘携带放射性物质，颗粒起尘会提高反应堆一回路粉尘浓度，改变放射性物质分布，在破口事故下颗粒起尘还会导致粉尘通过破口泄漏，从而给环境带来潜在的放射性污染。因此，研究湍流边界层内石墨粉尘的起尘现象，对高温堆放射性源项分析和安全评估具有重要意义。 石墨粉尘起尘现象的本质是沉积在壁面的颗粒在气流气动力和壁面粘附力作用下的动力学行为。针对已有研究未充分考虑高温堆环境下石墨粉尘形貌、壁面结构特性以及沉积结构对颗粒起尘影响等不足，本文基于湍流边界层和颗粒动力学理论，通过颗粒起尘的可视化实验，并结合计算流体力学、固体力学有限元等数值模拟技术，对湍流边界层内颗粒起尘特性开展研究，最终建立颗粒起尘的机理模型。 首先，本文对近壁面石墨粉尘受到气流的气动曳力开展研究，通过典型非球形石墨颗粒气动曳力数值计算，提出了描述颗粒形貌的无量纲参数，结合高温高压氦气氛围下颗粒表面滑移效应，建立了近壁面非球形石墨颗粒气动曳力的计算模型。 然后，本文对石墨粉尘受到粗糙壁面的粘附力开展研究，通过原子力显微镜测量了石墨粉尘颗粒与粗糙壁面的粘附作用，发现实验测量粘附力数值和经典粘附力模型预测结果存在量级上的偏差。为探究偏差产生原因，本文基于固体力学和分子间作用力理论，使用有限元法实现了颗粒-壁面接触粘附力的耦合模拟，证明了粗糙壁面导致的接触面积减少是粘附力偏差的主要原因。基于此，本文提出接触修正系数概念，并据此推导获得了适用于粗糙壁面-石墨颗粒粘附力的数学模型。 最后，本文搭建了石墨颗粒起尘的可视化实验平台，提出了基于图像识别的颗粒起尘份额检测方法，开展了石墨粉尘颗粒起尘的实验研究。随后，通过大涡模拟获得湍流边界层内脉动统计特征，结合本文建立的非球形石墨颗粒气动曳力模型和粗糙壁面石墨颗粒粘附力模型，建立了基于力矩特性的单层颗粒起尘机理模型。在单层颗粒起尘模型基础上，本文进一步发展了多层颗粒起尘的分区递推模型，该模型体现了颗粒沉积孔隙率及沉积厚度等对起尘的作用机理，模型预测和经典实验结果一致。基于该模型分析了高温堆工况一回路中石墨粉尘的起尘特性。

During the operation of the pebble-bed high temperature gas-cooled reactor (HTGR), the graphite-coated spherical fuel elements will generate graphite dust due to contact wear effects. When the reactor is operated under transient working conditions or breach accident occurence, the graphite dust deposited on the wall will resuspend from the wall under the action of the transient flow. Since the graphite dust carries radioactive substances, particle resuspension will increase the concentration of radioactive substances, and change the distribution of radioactive substances in the primary circuit. During the breach accident occurrence, particle resuspension leads to the leakage of radioactive dust through the breach, thereby bringing potential radioactive pollution to the environment. Therefore, it is important to study the resuspension phenomenon of graphite dust in the turbulent boundary layer for the analysis and safety assessment of the radioactive source term of nuclear reactors. The essence of the dust resuspension phenomenon is the dynamic behavior of the particles deposited on the wall under the action of airflow aerodynamic force and wall adhesion force. In view of the deficiencies in the existing studies that do not fully consider the graphite dust morphology, wall structure characteristics, and the influence of the deposition structure on particle resuspension, this thesis studied the characteristics of particle resuspension in the turbulent boundary layer through particle dynamics theory, resuspension visualization experiment, and numerical simulation technologies such as computational fluid dynamics (finite volume method) and solid mechanics (finite element method) to establish the mechanism model of particle resuspension. First of all, this thesis studied the aerodynamic drag of the near-wall graphite dust by the airflow. Through the numerical calculation of the aerodynamic drag of typical non-spherical graphite particles, a dimensionless parameter to describe the particle morphology is proposed. Combined with the particle surface slip effects due to high temperature and pressure helium, a calculation model of the aerodynamic drag of non-spherical graphite particles near the wall was established. Then, this thesis studied the adhesion interaction bewteen graphite dust and rough walls. The adhesion is measured by atomic force microscopy and it is found that there is an order of magnitude deviation between the experimental measured value and the classical adhesion model prediction. In order to explore the reasons for the deviation, this thesis uses the finite element method to realize the simulation of particle-wall contact adhesion based on the theory of solid mechanics and intermolecular forces. It is proved that the reduction of the contact area caused by the rough wall is the main reason of adhesion deviation. Therefore, the contact correction coefficient is proposed and then the adhesion mathematical model for rough wall contact is deduced accordingly. Finally, this thesis built a visual experimental system for graphite particle resuspension, proposed particle detection method based on image recognition, and conducted graphite dust resuspension experiment. Then, the statistical characteristics of fluctuation in the turbulent boundary layer were analyzed by large eddy simulation. Combined with the aerodynamic drag model and the adhesion force model, a torque-based monolayer particle resuspension mechanism model was established. Then, based on the monolayer particle resuspension model, this thesis further developed a partition recursive model for multi-layer particle resuspension, which reflects the mechanism of particle sedimentary porosity and sedimentary thickness on resuspension. The model prediction is consistent with the classical resuspension experiment results. Based on this model, the resuspension characteristics of graphite dust in the primary circuit of HTGR were analyzed.