【作者】 周国忠；

【导师】 施力田； 王英琛；

【作者基本信息】 北京化工大学 ， 化学工程， 2002， 博士

【摘要】 搅拌设备在化工、食品、冶金、造纸、石油和水处理等行业中的应用非常广泛。目前对其虽然已有许多的实验和理论研究，但有关的理论及设计计算方法仍不完善，许多还是处在半经验半理论阶段。近年来，利用激光多普勒（LDV）和计算流体力学（CFD）相结合的方法研究搅拌设备内的流动和混合特性逐渐发展起来。尤其是CFD技术，以其卓越的优势，对未来搅拌设备的研究、开发与设计所产生的影响不可估量。 本文首先利用商业CFD软件（CFX）提供的各种条件，对搅拌槽内的三维流动场进行了详细的基础数值研究，考察了桨叶区处理方法，湍流模型，网格数量和差分格式对计算结果的影响。通过对叶轮附近速度与湍流动能分布的研究发现，以上诸因素对速度分布的影响主要体现在最大径向、切向速度上。滑移网格法计算结果比多重参考系法略大；RNGκ-ε模型通过系数C₁^*引入描述流场畸变效应的附加源项后，其计算结果要优于标准κ-ε模型；随着网格数量的增加，计算结果更接近实验测量结果。在高密度网格下可以清楚观察到桨叶附近的尾涡，但计算得到的尾涡衰减很快；二阶迎风的计算结果较其它差分格式，包括三阶的QUICK格式的计算结果更加准确。对速度分布的预报并不是阶数越高越好。对于湍流动能，所有计算结果均较实验值偏低，增加网格数量，采用高阶的差分格式，如QUICK格式，能够提高对湍流动能的预报精度；RNGκ-ε模型虽然对速度分布的预报有所改善，但对湍流动能的预报却不如标准κ-ε模型。对搅拌槽内计算结果的最终改善需要采用新的方法才能从根本上提高计算结果的准确性。随着计算机计算能力的提高，大涡模拟和直接数值模拟方法用来研究搅拌槽内的流动场已逐步开展起来，并将逐步取代现行的方法，实现对计算结果准确性的提高。 在槽径为φ500mm的有机玻璃槽内，桨型为新型单层和双层轴向流CBY搅拌桨，采用激光多普勒测速仪测量了牛顿流体（水）和假塑性非牛顿流体（CMC水溶液）的流动场。研究了不同桨径，不同浓度非牛顿流体，以及多层搅拌桨不同层间距下的流动特性。针对实验研究的内容，利用商业CFD软件（FLUENT）对不同条件下非牛顿流体的湍流流动进行了数值模拟，并将计算结果与实验测量结果进行了对照。 实验结果结果表明：湍动状态下，牛顿流体与非牛顿流体宏观流动的差别主要在于CMC水溶液粘度增大而引起的主体流动减弱。叶轮区轴向速度较大，径向速度北京化工大学博士学位论文很小，其速度分布与环形射流速度分布类似。不同搅拌桨直径下，流动具有相似性，水与不同浓度CMC水溶液的轴向、径向速度分布基本一致。对双层搅拌桨体系，不同层间距下宏观流动形态不一样。上层桨与下层桨叶轮区、循环区速度分布与层间距密切相关。从流动场角度考虑，两桨之间存在最佳的层间距。 数值模拟结果较好地预测出了非牛顿流体湍流状态下的流动特征。流动场计算结果与测量结果基本吻合。对单层桨，叶轮区速度分布与实验结果基本吻合;循环区计算结果与实验数据尚有差距。对低浓度CMC水溶液，模拟结果较好:随着CMC水溶液浓度的增大，数值模拟结果精度略有降低。对双层桨叶轮区速度分布的计算结果与实验结果也存在一定差距。其精度没有对单层桨的计算结果准确，尤其是对径向速度的预测结果。 根据混合过程的原理，在CFX软件的基础上开发了混合过程计算程序，从CFD的角度对单层和多层涡轮搅拌桨在搅拌槽内的混合过程进行了研究。计算结果表明，速度场与浓度场联立求解与单独求解所得到的计算结果有一定区别，单独求解所得到的混合时间要比联立求解大，但是单独求解完全可以将问题表述清楚，同时其计算工作量小，计算比较灵活。以不同湍流模型的流动场计算的混合时间明显不同，表明混合过程与计算采用的流动场密切相关。混合时间的大小与加料位置和监测点的位置都有关系。在搅拌桨附近加料所得的混合时间最小，在槽底处加料混合时间最大:相同加料位置，监测点在槽底部时混合时间最大。对于多层桨的混合，当两个搅拌桨产生的流动能很好地连接起来时，混合非常迅速:当搅拌桨之间出现分区时，混合过程减慢。分区是控制整个搅拌槽内混合的关键因素。更多还原

【Abstract】 Mixing equipments are widely used in many industry processes. Althouth there are lots of theoretical and experimental study, the study of mixing and the design methods are not developed well up to now. The semi-empirical methods are still the main tools for design and scale-up industry processes. To meet the purpose for optimal design, scale-up and fully understand the processes occurring in agitated equipments, there are need for careful experimental measurements and mathematical models to measure and describe the characteristics of flow in such equipments. The combined method of using Laser Doppler Velocimetry (LDV) and Computational Fluid Dynamics (CFD) for the study of flow and mixing characteristics in stirred tank is carried out more and more frequently.In this study, based on the function provised by the commercial CFD code (CFX), three-dimensional flow fields in stirred tank are carefully studied, including the influence of impeller models, turbulent models, grid density and difference scheme to the computational results. The influence of such factors on velocity distribution near the impeller mainly focuses on the maxium radial and tangential velocity. It is found that the results using sliding grid, RNG k- e turbulent model and high-density grid are in good agreement with experimental data. While for difference scheme, the results of second-order upwind are better than that of QUICK. For turbulent kinetic energy, all kinds of calculation are under-predicted. With high-density grid and high order difference scheme, such as QUICK, the computation can get better results. Although RNG k- ?turbulent model is good for velocity distribution, its results of turbulent kinectic energy is worse than the standard k- ?model.The turbulent flow fields of Newtonian (water) and Non-newtonian (CMC water solution) fluids have been measured by using Laser Doppler Velocimetry. The research includes the influence of impeller diameter, concentration of CMC solution and interlayerspacing of dual impeller to the flow field. The experimental results show that the two different fluids have the same flow form under turbulence. The main difference is that the velocity of CMC solution is weakened because of high viscosity. The velocity distribution near the impeller, with high axial velocity and low radial velocity, is the same as ring jet flow. The flow form of dual impeller and the velocity distribution near the impeller depend highly on the interlayer spacing.According to the experimental research, the commercial CFD code (FLUENT) is used to simulate the turbulent non-Newtonian fluid flow in single and dual impeller stirred tank systems, and the results are compared with LDV measurements made in this dissertation. The predicted results are in good agreement with experimental research for single impeller flow field, and the lower concentration of CMC solution, the better of the predicted results. For velocity distribution of dual impeller systems, there is still deviation between the predicted and experimental results, especially for the predicted radial velocity.Program for mixing calculation was developed based on the commercial CFD code (CFX). It was used in the numerical study of mixing process in single and dual Rushton turbine stirred tank systems. Coupled and segregated solution of momentum and mass equation were adopted. The calculated concentration response curve is consistent with the literature data. Both methods predict the same change rule of mixing time. But the value of mixing time from coupled solution is lower than that from segregated one. Coupled solution needs much more computational efforts than the segregated. It has shown that mixing process depends on the flow field used for mixing calculation. The value of mixing time also depends on detecting and feeding location. When the tracer is fed near the impeller, the mixing time is the lowest. While it is fed near the bottom of the tank, the mixing time is the largest. For dual impeller stirred tank system, when the flow generated by the two impellers has更多还原