流化床中的多尺度非均匀结构对两相的流动、传热、传质和反应具有重要影响。传统的双流体模型（TFM）结合均匀的颗粒群曳力本构关系，忽视了网格内的非均匀结构，无法准确预测流化床中的流动特征和反应、传递行为。为了改进流化床的数值模拟，需要建立考虑网格内结构的介尺度模型。其中，基于多尺度结构分解以及稳定性条件假设的能量最小多尺度（EMMS）模型，能有效解析流化床中的稀密两相结构分布，并准确预测循环流化床中的流域转变；而基于EMMS的曳力模型，亦可准确描述亚网格结构导致的曳力下降，因而在流化床的数值模拟中得到了广泛应用。现有EMMS曳力是在稳态EMMS模型基础上，引入颗粒相加速度来建立非稳态的亚网格模型。然而，这些额外引入的动态变量，与稳态EMMS模型中既有的稳定性条件是否相容，尚未得到充分证明；而诸多基于极值型原理（包括非平衡热力学）的稳定性条件在流态化系统中的适用性问题也有待进一步分析和研究。此外，在算法上，现有EMMS曳力模型求解皆依赖于床层的整体操作条件(Ug,Gs)，不同流域条件下需要重新计算拟合，缺乏通用性。基于上述问题，本论文在EMMS方法基础上，发展了新的基于稳态假设的、且对不同流域普适的非均匀亚网格曳力模型；针对稳定性条件的适用性，开展了流态化系统的能量耗散率极值分析，指出EMMS稳定性条件的优势以及传统的最小耗散原理的不足；进一步，根据两相脉动能和耗散率的分析，提出了新的聚团模型，将EMMS预测的流域转变从噎塞拓展到最小鼓泡，扩展了EMMS方法的适用范围。具体如下：首先，论文第二章基于稳态EMMS模型提出了具有流域普适性的非均匀曳力模型。此模型可以同时适用于鼓泡床、湍动床和循环床等多个流域。相较于之前版本的EMMS曳力模型，新的曳力模型关系式不需要随流化床的流域或操作条件的改变而重新拟合曳力关系式。此外，基于双流体模型的模拟验证发现，新的曳力模型能合理捕捉到鼓泡床、湍动床和循环床中的流动特征。其次，基于稀密两相二元分布的假设，通过空间平均方法，论文第三章推导了结构多流体模型（SFM）的质量、动量、能量和熵平衡方程。其中质量和动量平衡方程能有效阐明SFM与TFM和EMMS模型的平衡方程的关系。能量平衡和熵平衡方程为进一步分析流化床中的能量传递和耗散过程，以及耗散率极值原理在流化床中的适用性奠定了基础。基于上述SFM模型的平衡方程，第四章推导了基于结构的熵产率和能量耗散率的关系式，并探索了能量耗散率极值原理在气固流态化系统中的适用性。结果显示，在考察的流域范围内，能量耗散率极小原理适用于理想输送的流域，能量耗散率极大原理在密相区与EMMS的稳定性条件的预测一致，而EMMS的稳定性条件能合理预测“噎塞”流域转变现象。基于SFM模型的平衡方程，第五章推导了稀密相之间的介尺度脉动能的表达式，并且通过量纲分析，建立了聚团尺度与介尺度脉动能和能耗率的关联式。将新的聚团模型代入到EMMS模型中求解，可以准确地预测鼓泡床的床层膨胀曲线，特别是Geldart A类颗粒和 B类颗粒在最小鼓泡点附近不同的流域转变现象。论文最后总结了所得到的主要结论，并对模型的前景以及进一步研究的方向进行了展望。;Multi-scale heterogeneous structures have significant effects on the hydrodynamics, heat transfer, mass transfer and chemical reactions in fluidized beds. Traditional two-fluid model (TFM), integrated with homogeneous drag force, fails to predict the hydrodynamic characteristics, reaction and transport behaviors in fluidized beds, because it neglects sub-grid heterogeneous structures. In order to improve the numerical simulation of fluidized beds, it is necessary to take into account the sub-grid heterogeneous structures in the drag model. The energy-minimization multi-scale (EMMS) model, which is based on the multiscale resolution of structure and closed with a stability condition, enables effective solving the dense-and-dilute two-phase distribution in fluidized beds and predicting the regime transitions in circulating fluidized beds. Meanwhile, the EMMS-based drag model can effectively quantify the sub-grid drag reduction induced by heterogeneous structures, and hence has been widely applied in numerical simulations of fluidized beds. The previous versions of EMMS drag models introduced the particle acceleration terms to account for the unsteady state behavior. However, the consistency between these extra dynamic variables and the stability conditions of the steady-state EMMS model has not been proven. Moreover, the validity of various extremum principles, including nonequilibrium thermodynamics, needs to be analyzed in the field of fluidization. Further, as to algorithm and solution, current EMMS drag models heavily depend on the operating conditions, thus need case-specific regression and calculation, with extra expense of computation, and lack universality in application. To resolve the above issues, firstly, this thesis aims to develop a drag model suitable for more flow regimes based on the steady-state EMMS framework. Secondly, to understand the applicability of stability conditions, the extremum analysis of energy dissipation rate is carried out for fluidized beds, indicating the advantage of the EMMS stability condition and the problem of using the minimum energy dissipation rate principle. Further, a new model for cluster diameter is proposed based on the analysis of two-phase fluctuation energy and dissipation, and thereby enables prediction of the minimum bubbling state beside the choking transition. The application of EMMS method is thus extended. In Chapter 2, an EMMS-based heterogeneous drag model is proposed based on steady state EMMS model. The new drag model without tunable parameters is suitable for multiple flow regimes, such as bubbling fluidized beds, turbulent fluidized beds and circulating fluidized beds. Compared with the previous versions of EMMS-based drag model, the constitutive relations of this new drag model is independent of regimes or operating conditions. Two-fluid modeling, integrated with the new drag model, indicates that it can reasonably predict the hydrodynamic characteristics of bubbling fluidized beds, turbulent fluidized beds and circulating fluidized beds.Then, based on the assumption of dense-and-dilute two-phase distribution in fluidized beds, the mass, momentum, energy and entropy balance equations of the structure-dependent multi-fluid model (SFM) are derived by applying the volume averaging method in Chapter 3. The mass and momentum balance equations of SFM help clarify the relationships among SFM, TFM and EMMS balance equations. The energy and entropy balance equations pave the way for analyzing the process of energy transport and dissipation, and the extremum behavior of energy dissipation rate in fluidized beds.In Chapter 4, the structure-dependent entropy production and energy dissipation rate of SFM are derived from the balance equations of SFM, and further, the extremum analysis of energy dissipation rate is carried out. The results show that the minimization principle of energy dissipation rate is applicable to ideally dilute pneumatic conveying. The result predicted by the maximization principle of energy dissipation rate is consistent with that predicted by EMMS stability condition in the dense flow region. Specifically, the EMMS stability condition predicts the choking transition phenomenon.In Chapter 5, the mesoscale fluctuation energy between the dense and dilute phases is derived based on SFM balance equations. The length scale of clusters is correlated with mesoscale fluctuation energy and its dissipation rate by applying the dimensional analysis method. By substituting this new correlation into EMMS model, it effectively predicts the bed expansions in bubbling fluidized bed and the different regime transition phenomena near the minimum bubbling of both Geldart groups A and B particles.Finally conclusions and perspectives are given.