低碳烯烃（如乙烯和丙烯）是化学工业中重要的有机化工原料。传统工艺以石脑油为原料，通过蒸汽裂解和催化裂化副产工艺进行生产。我国的煤炭储量丰富，随着石油资源的日益枯竭，近年来，以煤为原料生产低碳烯烃的MTO/MTP（甲醇制烯烃/甲醇制丙烯）工艺得到广泛关注。然而，目前大多数的研究工作集中于MTO/MTP催化剂的合成、制备和设计，对工业反应器内部的流动和反应行为的研究较少，因此，MTO/MTP工业反应器的放大仍然困难。近年来，计算流体力学(CFD)发展迅速，在反应器放大这一领域已有较多应用。然而，MTO/MTP工业反应器中的焦炭生成速率慢、催化剂停留时间长、反应器尺寸大且处于湍动流化区域。这些特点对CFD模拟应用提出了新的挑战，比如：由于焦炭生成速率慢和停留时间长，如果从新鲜催化剂开始CFD模拟，需花费数百天的时间才能达到稳定，因此，如何加速CFD计算非常重要；对于湍动流化床，现有的EMMS/bubbling曳力模型能否适用，尚需检验；对于工业反应器，受限于计算量，需采用粗网格计算，因此有必要发展对网格大小不敏感的粗网格计算模型。针对上述问题，本文开展了以下研究。论文第二章提出化学反应工程(CRE)模型与CFD耦合的计算方法：将CRE预测的焦炭浓度赋给CFD模拟作为初场值，进而通过多尺度CFD进行动态演化。通过MTO中试反应器验证，结果表明，基于鼓泡床特点建立的CSTR(Continuous Stirred Tank Reactor)模型与CFD的耦合可以大大缩短从初始态到稳态的过渡时间。在此基础上，还讨论分析了反应器中催化剂浓度、速度、组分浓度的分布和考察了不同的反应动力学模型。论文第三章将该EMMS/matrix的二步法思路扩展应用于EMMS/bubbling曳力模型。使EMMS/bubbling曳力的非均匀因子扩展为局部空隙率和滑移速度的二元函数。进一步，通过曲面拟合的方法，实现与商业CFD软件的耦合。针对多个流化床的模拟结果表明：二步法EMMS/bubbling曳力能准确捕捉湍动床内的浓稀两相结构，而且相对原有曳力模型，其准确性更高，网格依赖性也更低。基于上述工作，论文第四章采用二步法EMMS/bubbling曳力模型以及CRE确定的初场分布，实现了DMTO工业反应器的快速模拟。结果显示，曳力模拟预测的流动参数（如压降、平均固相空隙率）与实验值或经验值吻合较好，说明该方法可用于粗网格模拟。另一方面，预测的乙烯和丙烯与实验值的偏差较大，这可能需对实验室装置上获得的反应动力学模型进一步改进，以体现反应器放大过程中带来的流动结构、催化剂停留时间分布等一系列因素的影响。论文最后对全文进行了总结，并对本文提出的模型应用前景以及MTO反应器模拟进行了展望。

Light olefins such as ethylene and propylene are important petrochemicals，from which a number of downstream products can be produced. The conventional methods of producing olefins mainly include naphtha steam cracking, and catalytic cracking processes. China has rich coal resources but is poor in oil and natural gas reserves. So recently, the MTO/MTP (Methanol to olefins/Methanol to Propylene) processes whose raw material (i.e. methanol) can be obtained from both oil and non-oil resources, such as coal and natural gas, are considered to be economical routes to produce light olefins and thus have attracted much attention. However, previous studies mainly concentrate on catalyst design and chemical reaction kinetics. The study about multiphase flow and reaction behaviors in MTO/MTP fluidized bed reactor are rarely reported. Therefore, it is very difficult to directly scale up a MTO/MTP reactor from laboratory-scale or pilot-scale to industrial-scale. Recently, the computational fluid dynamics(CFD) is developing very fast and it can be used to design and scale-up a reactor. However, in MTO/MTP reactor, it is low coke generation rate and long residence time, operates in turbulent fluidization, and large scale. Those features pose new challenges to current CFD modeling. Firstly, owing to low generation rate of coke and long residence time, it normally takes over a hundred days to reach the desired coke content if one starts CFD simulation from fresh catalysts, which is obviously unacceptable in practice. Secondly, it is questionable whether the current EMMS/bubbling drag model can be applied in the turbulent fluidization. Finally, coarse-grid simulation for an industrial MTO reactor requires development of nearly grid-independent approach. To resolve these issues, this study will be conducted on the following aspects.In Chapter 2, we apply the Chemical Reaction Engineering (CRE) modeling to predict the coke content and then take it into CFD as the initial state value. This combined approach is tested in simulating a DMTO pilot-scale reactor. And it can greatly shorten the transition time from the initial state to the steady state. In Chapter 3, we extend the two-step scheme of EMMS/matrix model to the EMMS/bubbling model. The heterogeneity index of drag coefficient is found to depend on both local gas volume fraction and slip velocity. The fitting functions of the heterogeneity index are then integrated into CFD solver. Simulation of turbulent fluidized beds and a MTP pilot-scale reactor with this two-step EMMS/bubbling drag model shows reasonable agreement with experiments and weaker dependence on grid resolution.Simulation of an industrial-scale DMTO reactor is then conducted in Chapter 4, based on the testing and validation in Chapter 2&3. It is found that the prediction of pressure drop profiles and solid inventory agrees well with experimental data. The difference between the predicted mass fractions of ethylene and propylene and the experimental data is bigger than the case of the pilot-scale reactor. That implies chemical kinetics based on bench-scale beds may not be used for general purpose. And the scale-up effects such as the change of flow structure and RTD should be considered in future work.In Chapter 5, we summarize the whole article with prospects and suggestions on MTO/MTP reactor simulation