CAS OpenIR  > 研究所(批量导入)
基于光滑粒子动力学的双流体模拟
Alternative TitleTwo-fluid smoothed particle hydrodynamics for particle-fluid two-phase flow
邓利娟
Subtype博士
Thesis Advisor李静海
2013-05-01
Degree Grantor中国科学院研究生院
Degree Discipline化学工程
Keyword气固两相流   光滑粒子动力学   双流体模型   gpu编程   介尺度
Abstract气-固两相流的计算流体力学模拟常用两大类模型:连续介质模型和离散粒子模型,前者通常采用欧拉坐标系下的基于网格的算法,算法成熟稳定,计算量相对较小,但模型难于处理复杂边界,而且流体区域之间耦合性较强,并行效率不高;后者则多采用拉格朗日坐标下的由时间或事件驱动的无网格算法,该算法得到的颗粒流场局部特征比较明显,并且适合处理复杂边界、适合并行计算,但这种算法计算量很大,难以满足大规模工业装置的模拟需要。本论文的研究目标是采用无网格的光滑粒子动力学(SPH)算法求解连续介质模型,综合两者的优点,希望实现工业规模装置的快速、准确计算。论文首先建立了双流体模型(TFM)的完全的光滑粒子动力学(SPH)求解框架。该模型将宏观连续的流体及宏观离散的颗粒皆处理成SPH离散粒子。总结了常见的几种边界条件在SPH中的实现形式。遵循由简至繁的原则,首先针对单相液体Couette流分析了SPH弱可压缩假定中人工声速的影响。发现增大人工声速会增大系统的能量损耗,从而影响计算结果。然后用两相SPH模型中的固相的控制方程模拟了颗粒流的Couette流动,发现SPH的模拟结果与理论解相吻合,而且颗粒流处理成连续流体后,颗粒的直径对模拟结果影响很小。本章最后,通过对液-固方腔剪切流、液固沉降和气固鼓泡床的模拟,对两相SPH模型进行了定性和定量验证。发现SPH模型可定量模拟液-固剪切现象、液固沉降问题和鼓泡流化现象,但计算精度仍有待于进一步的提高。鉴于计算规模增大对计算效率的要求,论文第三章建立了两相流的完全SPH模型的CPU、GPU和multiGPUs算法。通过实际测算指出:计算规模达到一定程度以后,GPU相对CPU的计算优势才更加明显;在选择multiGPUs算法时,应尽量降低MPI通信所占的时间才能充分发挥多块GPU卡并行的计算优势。考虑到SPH中固有的弱可压缩假定,完全的SPH难以得到令人满意的压力场。因此,论文第四章建立了基于TFM的欧拉(气相)—拉格朗日(固相)混合模型——TF-SPH模型。该模型对气相采用欧拉坐标下的差分方法求解,固相采用SPH离散,并通过EMMS模型考虑介尺度结构。经过一系列参数敏感性分析后,通过模拟气固提升管对模型进行了定量验证。现有的初步结果显示:在部分操作条件下,TF-SPH模型的计算结果与实验数据相吻合,但对更宽的操作条件下的适应性,还有待深入研究和改进。论文最后总结了本论文得到的主要结论和创新点,并提出了下一步的研究方向。
Other AbstractComputational Fluid Dynamics (CFD) for gas-solid two-phase flow follows two main strategies: the continuum approach and discrete particle approach. The continuum approach is usually solved by using mesh-based algorithm on Eulerian coordinates, whose computational load is relatively low. It is, however, not suitable for solving problems with complicated complex geometric boundaries and is inefficient to parallel computing. The discrete particle approach is usually solved by time-driven or event-driven meshless algorithm on Lagrangian coordinates and is suitable for parallel computing. However, it is computation-intensive and thus impractical to be used for large-scale industrial simulation. In this thesis, we aim to combine the merits of both the continuum model and the meshless, smoothed particle hydrodynamics (SPH) algorithm to realize accurate and rapid simulation of fluidized beds. First, we presented the SPH for solving two-fluid model (TFM). To validate this method, a liquid couette flow was simulated to analyze the effects of the artificial sound speed. We found that larger sound speed would result in larger energy dissipation. Then we simulated the granular couette flow by swithing off the interaction force and the pressure drop in TFM. It was found that the SPH solution agrees well with theory and the particle diameter has little effect on the final result when particles are treated as a continuum. In the final part of this chapter, we simulated a shear driven cavity flow of liquid-solid mixture, a liquid-solid sedimentation and a gas-solid bubbling bed to validate the model. It was found that the SPH simulations are in quantitative agreement with reality.The coding detail for CPU, GPU and multiGPUs systems were then presented. It was found that the speedup ratio of GPU computing was remarkably high when the computation scale is big enough. The multiGPUs algorithm is viable only when MPI communication ratio is low enough. To avoid the errors induced by the strong compressibility in TFM, we proposed a TF-SPH model which coupled the Eulerian solution of the gas phase and the Lagrangian solution of the solid phase (chapter 4). The EMMS model was used for the meso-scale structure. A series of sensitivity analysis were made in order to choose the proper SPH parameters. Quantitative agreements can be found in simulation of a circulating fluidized bed.Finally, we summarize the main achievements in this work, and then present perspectives on the TF-SPH methods.
Pages137
Language中文
Document Type学位论文
Identifierhttp://ir.ipe.ac.cn/handle/122111/8250
Collection研究所(批量导入)
Recommended Citation
GB/T 7714
邓利娟. 基于光滑粒子动力学的双流体模拟[D]. 中国科学院研究生院,2013.
Files in This Item:
File Name/Size DocType Version Access License
基于光滑粒子动力学的双流体模拟.pdf(9004KB) 限制开放CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[邓利娟]'s Articles
Baidu academic
Similar articles in Baidu academic
[邓利娟]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[邓利娟]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.