颗粒流体系统广泛地存在于自然界和工业生产中。由于微尺度上存在固固间的耗散作用及气固间的非线性作用、介尺度上存在颗粒控制机制与流体控制机制在竞争中的协调作用，以及宏尺度上边界对流动特性的约束，导致两相流系统中有明显的非均匀结构产生。这些非均匀结构对系统内质量、动量、能量的传递及内部发生的化学反应均有十分显著的影响。掌握气固两相流系统的流动控制机制，探索非均匀结构的产生机理，量化非均匀结构对“三传一反”的影响，对工业放大和反应器设计均具有十分重要的意义。随着近些年来计算机技术的迅猛发展，基于计算流体力学的模拟方法被广泛地应用于探索非均匀气固两相流系统。目前，连续介质模型在工业设备的模拟中有着较为广泛的应用。但由于气固两相流系统中非均匀结构的普遍存在，在用细网格模拟时在结构交界处局部浓度、速度、温度的梯度较大，所以基于局部热力学平衡假设的Navier-Stokes 阶的连续介质模型是否适用还需进一步考察；另外，在用粗网格模拟时，网格内非均匀结构的存在是系统中不同的流动控制机制在竞争中协调的结果，不同流动控制机制导致有颗粒主导的密相及气体主导的稀相的产生，竞争中协调这一原理如何用于改进现有连续介质力学也需进一步探索。本博士论文围绕这两个关键问题展开，主要开展了以下工作：本博士论文的主体工作是采用非平衡态统计力学分析非均匀气固两相流，特别是充分利用了颗粒动理学和相关方法，因此本论文第一章系统阐述了颗粒动理学的研究进展。现在广泛使用的连续介质模型是建立在局部热力学平衡假设的基础上的，但是由于局部热力学平衡假设仅对于线性非平衡态系统成立，而基于局部热力学平衡假设的连续介质模型无法定量的给出此偏差的数值范围，所以论文第二章就是从动理论的角度来确定局部热力学平衡假设成立的定量标准——熵判据。此判据定量的刻画了非线性项的影响，确定可保证局部热力学平衡假设成立的非线性项的范围，进而为Navier-Stokes阶的连续介质模型成立提供了定量的判断依据。此判据不仅与颗粒温度梯度和速度梯度相关，也与颗粒间的非弹性碰撞性质密切相关，所以此熵判据相较于基于努森数的经典判据更为全面。在第二章中已初步导出了气固两相流中熵守恒方程的形式，但其涉及到的本构关系还未封闭。论文的第三章进一步用集团展开法考虑颗粒间位置的关联性，改变径向分布函数形式并将其代入传统Enskog方程中，最后基于修正的Enskog方程，结合Chapman-Enskog（C-E）方法导出更为合理的熵方程，同时也封闭熵通量及熵产率。由熵产率分析气固两相流系统是否失稳，从统计力学的观点来阐述介尺度结构的形成机理。论文的第四章是为已有的稳约双流体模型构建统计力学基础：其中密相部分，拓展Boltzmann动理论，成功建立可考虑气固两相流中颗粒聚团体积、密度和速度时空动态变化的动理论，为受稳定性条件约束的连续介质模型奠定了统计力学基础；与此同时，因兼顾考虑分子尺度和宏观尺度的速度脉动，在国际上首次通过底层动理论直接导出气固两相流中合理的气相控制方程，其中同时包含分子应力项和拟雷诺应力项，而此前的研究中由于忽略不均匀结构的影响，拟雷诺应力项则不会出现在气相动量守恒方程中。然后通过计算流体力学模拟验证了此模型的合理性并初步考察了气相拟湍流作用的影响。论文的第五章借助于模拟计算中改变几何结构的便利性，用稳约双流体模型系统研究了高密度循环流化床出口几何结构的影响。得到的结论如下：（i）出口类型影响较为显著，将弱约束出口改为强约束出口时不仅会使靠近出口处固含率增大还会使得整个床层上部区域均有变浓的趋势；（ii）改变强约束出口空腔高度、弱约束出口曲率半径及出口延伸管长度对床内流动几乎无影响；（iii）缩小强约束出口的出口管径可明显增加床内存料量，同时增强在接近出口区域处的固含率径向分布的对称性，而对弱约束出口而言缩小出口管径对模拟结果几乎无影响。以上模拟结果与文献中的试验结果定性完全一致，这也进一步说明了稳约双流体模型模拟非均匀气固两相流的合理性。第六章对本论文进行总结，总结了本论文的主要结论及创新点并对下一步的研究方向进行了展望。;Particle-fluid system can be widely founded in nature and in engineering process. The dissipative particle-particle collisions and nonlinear interactions between gas and particles at microcale, the compromise in competition between gas-dominated and particle-dominated mechanisms and the constraints of boundary conditions all result in the formationof multiscale spatio-temporal heterogeneous structures. These complex structures have a significant influence on mass, momentum, energy transport and chemical reaction in the system. With the rapid development of computer technology, computational fluid dynamics (CFD) has been extensively used to study gas-solid flow.Continuum model has been chosen frequently for industrial applications. However, solid concentration gradient, velocity gradient and granular temperature gradient are usually high at the boundaries of heterogeneous structures. Hence, it's necessary to assess whether the Navier-Stokes order continuum model on the basic of the local thermodynamics equilibrium hypothesis is suitable for the system or not; moreover, different dominant mechanisms have to compromise each other at mesoscale which result in the formation of dynamic mesoscale structures and the coexistence of dilute phase and dense phase. Therefore, the issue about the effect of different dominant mechanisms need to be explored deeply.Chapter 1 introduces the analysis of kinetic theory, and the development of various kinetic theories has been reviewed in detail. In chapter 2, kinetic theory has been used to analyse the entropy of gas-solid systems, it was shown that local thermodynamic equilibrium postulate is only valid for linear non-equilibrium regime, furthermore, an entropy criterion for the boundary of the validity of Navier-Stokes order continuum models, which is the ratio of the second order correction of the entropy density to the equilibrium one and is the indication of the effect of non-linearity, can be obtained from the analysis.The basic form of entropy equation of gas-solid flow has been derived in chapter 2, but the constitutive relations have not been obtained yet. In chapter 3,The classic cluster expansion method is introduced to consider the correlation of granular positions, then the modified Enskog equation is established by submitting a modified pair correlation function into the Enskog equation. Furthermore, a more reasonable entropy equation and constitutive relation are deduced using the Chapman-Enskog expansion method. It was found that there is no H-theorem in dissipative granular gases and gas-solid flow, therefore, the instability will results in the formation of mesocale structures. In response to the heterogeneous gas-solid flow, an EMMS-based two fluid model has been developed from the viewpoint of continuum mechanics, however, its microscopic foundation remains unknown. In chapter 4, in case of particle phase, the generalized Boltzmann kinetic theory considering the spatio-temporal variation of the volume, density and velocity of clusters was firstly derived, which launches a solid microscopic foundation of EMMS-based two-fluid model; with respect to the gas phase, a new method was developed by considering the fluctuations at different scales simultaneously, with which we can for the first time derive the correct governing equations of gas phase via kinetic theory, in the sense that both the molecular stress and the Reynolds (or pseudo-Reynolds) stress can be obtained simultaneously, whereas all previous kinetic theory analyses failed to predict the appearance of Reynolds (or pseudo-Reynolds) stress in the momentum conservation equation of gas phase due to the assumption of uniform structure. Finally, CFD simulations have been carried out to validate the EMMS-based two-fluid model and to study the effect of gas phase pseudo-turbulence.It has long been recognized that the geometry of CFB risers has a significant effect on the bed hydrodynamics. However, a systematical study of exit effect is still lack after extensive experimental studies, possibly due to the fact that systematical modification of exit geometry in experimental study is an expensive and time-consuming task. In chapter 5, the EMMS-based TFM is used to systematically investigate the effect of exit geometry of high-density CFB risers, fully taking the advantage that bed geometry can be easily modified in computational fluid dynamics study. It is shown that (i) the type of exit has a significant effect on hydrodynamics, the use of abrupt exit results in an increased solids concentration, not only in the immediate vicinity of the exit but also for a considerable distance down to the riser; (ii) the cavity height of abrupt exit, the curvature diameter of smooth exit and the length of the horizontal tube connecting the exit and the primary cyclone only have a minor effect or have no influence on the bed hydrodynamics of the studied risers; (iii) more importantly, the decrease of the diameter of abrupt exit tube results in a remarkable increase of solids holdup, the symmetry of radial solids concentration near the exit can also be enhanced significantly. However, in case of smooth exit,the diameter of the exit tube has no effect on the bed hydrodynamics at all. All of those results are in agreement with conclusions obtained from previous experimental studies, thus offering further validation of the EMMS-based two-fluid model for modelling heterogeneous gas–solid flow.Finally, the main conclusions and innovations of the dissertation have been summarized in chapter 6, moreover, the perspective of future research are also presented.