流体在纳米通道中的限域传质现象广泛存在于现代化工技术。因缺乏对传质机制及调控方法的认识，人们难以对限域传质进行定量的描述，这使得相关材料及结构的应用受到制约。在限域传质中，传统流体力学模型普遍存在严重偏差，无法准确预测流动通量。此外，其突出的壁面作用对流动的影响十分显著，这使得限域传质的受控因素更具特殊性。以上种种问题都对纳米流体力学理论的发展提出了更进一步的要求。基于此，本文采用计算机模拟方法,分别针对纳米受限通道中的流体结构、压力驱动流动、毛细浸入流动及复杂大分子接枝壁面间的流动问题进行了系统的研究，探究了流体流动规律，发展了相关数学模型，同时在现有模拟方法的基础上开发了更为高效精确的多尺度耦合模拟方法。具体工作以及主要成果如下：（1）采用耗散粒子动力学（Dissipative Particle Dynamics, DPD）模拟方法对流体的纳米受限行为进行了深入的研究，考察了受限流体的非均匀结构及其特殊的流动现象。同时，利用DPD模拟研究了纳米受限通道的阻力构成，并探讨了孔道结构及壁面性质对流动阻力的影响。结果表明，纳米受限流体存在非均匀边界区，且边界区受壁面影响显著。边界区内的密度变化导致粘度变化，从而直接造成流速分布的改变以及通量的提高。此外，通过对纳米通道进行阻力分析，研究了孔道结构及壁面性质对流动阻力影响，并最终设计了阔口（亲水）-窄口（疏水）的优化结构。（2）基于Stokes方程，并综合考虑壁面附近流体的非均匀性，推导了适用于纳米受限流体的通量模型以及表观滑移距离模型，并采用DPD模拟结果以及前人的实验数据验证模型的准确性。在通量计算方面，推导建立了特征尺寸从原子尺度到宏观尺度的Poiseuille流体通量计算模型，打破了宏观流体与纳米流体之间理论的不互通性，并在理论上解释了纳米受限流体通量高于传统模型预测值的现象。在表观滑移距离方面，分别推导了适用于Couette流和Poiseuille流的表观滑移距离模型，该模型明确体现了表观滑移距离随通道宽度减小而增大的趋势。（3）基于水在层状石墨烯膜及碳纳米管膜中的毛细浸入及蒸发过程，采用分子动力学（Molecular Dynamics, MD）模拟方法研究了流体的自发输运。模拟研究表明，水在通道内的毛细浸入通量受表面张力、滑移距离、受限流体密度等多种因素影响，随通道宽度呈现非单调的变化过程，存在最优通道宽度。结合模拟结果，对描述毛细过程的Lucas-Washburn模型进行了修正，使其能够更为准确地描述纳米通道中的毛细过程及其受多种因素影响而产生的复杂变化趋势。而由毛细过程和蒸发过程合作产生的自发流动，随纳米通道长度的增加，流体通量由蒸发过程控制转化为毛细过程控制。针对蒸发控制过程和毛细控制过程，分别采用外、内表面亲水基团修饰可显著提高通量。其中，内表面的浸润性梯度使通道呈现流体二极管的特点。（4）基于现有的MD与DPD模拟方法，开发了MD-DPD耦合方法以在满足局部精确性的同时又能兼顾整体计算的高效性，并在简单流场中得到验证。该算法用于探索接枝聚合物及生物大分子的壁面与流场间的相互影响。结果表明，MD-DPD方法在解决复杂壁面间流体流动问题时具有明显的优势，其一方面能够达到与MD模拟相同的准确性，另一方面又在计算效率上实现了数倍至数百倍的提高。利用该方法进一步发展了大分子接枝壁面间的流体流动模型，可分别通过滑移距离修饰以及联立求解Brinkman与Stokes方程两种方式实现流场的求解。 ;With the recent development of micro-chemical technology and membrane science, the nanoflows become ubiquitous in chemical engineering. However, theoretical understanding about the nanoflows is still scarce when facing with the failure of traditional hydrodynamics. The missing of theoretical models hinders the application of nanoflows. Thus, further development of nanofluidics is being required to solve the problems such as the deviation between nanoflows and traditional theories, and the regulation of nanoflows. This work is aimed at providing a systematic investigation of nanofluidics, targeting at the confined fluidic structure, the pressure-driven flow mechanism, the spontaneous flow mechanism and the flow within grafted- macromolecules. It is expected to unravel the nanofluidic mechanisms, improve the theoretical models and develop multi-scale simulation methods with both accuracy and efficiency. The primary findings are listed as follows.(1) DPD simulation is employed to investigate the nano-confined behaviors of fluid, especially its heterogeneous structure and enhanced transport. It is demonstrated that nanoflows are closely correlated to the heterogeneous boundary region. Resulted from the heterogeneous density, the heterogeneous viscosity directly lead to a distortion of velocity profile and an enhancement in transport flux. Meanwhile, the resistance composition of a nano-channel is also analysed by DPD simulation. A funnel(hydrophilic)-channel(hydrophobic) structure is designed to optimize the flow resistance.(2) Involving the heterogeneity of nanoflows, theoretical models are derived for the transport flux and apparent slip length. The flux model is verified by DPD simulation and previous experiments. It is demonstrated that the model, on one hand, can be applied to nanofluidic systems to make precise predictions, and on the other hand, can be simplified to traditional hydrodynamic models when enlarging the channel width. The new model sheds a light on the nanofluidic mechanism and bridge the gap between the nano- and macro-fluidics. The model for apparent slip clarifies the increasing tendency of slip length with descending channel width.(3) Based on the membrane of layered graphene and carbon nanotube, and on the capillary and evaporation processes of water within the membrane, MD simulation is employed to study the spontaneous water transport in nanochannels. The capillary imbibition flux of water shows non-monotonic variation with the channel width, which results from a complex competition among surface tension, slip length, confined density, etc. Thus, an optimal width can be assigned to a nanoscale capillary. The Lucas-Washburn equation that is widely used for capillary flows is modified to correctly reproduce the water imbibition within nanochannels. For the capillary-evaporation process, the spontaneous water flow is subjected to different controlling steps according to the channel length. In short channels, the process is controlled by the evaporation, which can be accelerated by hydrophilic modification on the external surface. In long channels, however, the process is controlled by the capillary process, which can be promoted by hydrophilic modification on the internal surface. Besides, the hydrophilicity gradient on the internal surface of nanochannels can act as a fluidic diode. (4) A coupled MD-DPD multiscale simulation method is developed to achieve accurate and efficient calculation. The MD-DPD method is verified in simple flows, where it provides accurate Poiseuille and Couette flow fields compared with analytical solutions. When applied to flow with bio-molecule or polymer grafted surfaces, the method can reproduce the results from MD simulation while significantly reduces the simulation time. With the aid of the MD-DPD method, the flow within grafted macro-molecules is investigated. Two kinds of models can be used to depict such flows: one is the slip length model and the other one is the coupled Brinkman-Stokes model.