Multiscale Simulations of Heat Transfer and Fluid Flow Problems
Multiscale Simulations of Heat Transfer and Fluid Flow Problems
复制标题
DOI：
10.1115/ihtc14-23408
复制
发表时间：
2012-03
期刊：
影响因子：
--
通讯作者：
W. Tao;Ya-Ling He
中科院分区：
文献类型：
--
作者：
W. Tao;Ya-Ling He
多尺度模拟是一个快速发展的研究领域，将对工程计算数学和数值模拟产生重大影响。在这个主题演讲中包括以下部分。首先，什么是多尺度问题。在热流体科学中，多尺度问题可分为两类：多尺度过程和多尺度系统。通过多尺度过程，我们的意思是总体行为是由发生在不同长度尺度上的过程所控制的。我们所说的多尺度系统是指以长度尺度变化大为特征的系统。电子系统的冷却就是这样一个典型的多尺度系统。简要介绍了现有的三种几何尺度（宏观、中观和微观）的数值方法。第二部分讨论了多尺度模拟的必要性。给出了多尺度过程和多尺度系统的实例。本讲座的重点是多尺度过程的模拟。在第三节中，提出了用于模拟多尺度过程的数值方法。有两种类型的模拟方法。一种是利用一般控制方程，求解涉及特征几何尺度数阶变化的整个流场。另一种是所谓的“区域求解和界面耦合”。该方法采用不同的数值方法模拟不同长度层次的过程，然后在不同区域之间的接口处交换信息。信息交换应该以物理上有意义、数学上稳定和计算效率高的方式进行。关键是建立重构算子，将宏观计算的少量变量数据转化为微尺度或中尺度模拟的大量变量数据。针对不同的耦合情况，简要回顾了现有的求解算子的方法。第四部分给出了四个多尺度的数值模拟实例：基于MDS和FVM的粗糙纳米通道内的液体流动、基于DSMC的流体流动和基于FVM的固体流动的微喷嘴内的流动和换热、基于FVM和LBM耦合的圆柱体流动和方形腔内的自然对流换热。最后指出，要对PEMFC内输运过程和强化表面上制冷剂冷凝过程等复杂工程问题进行成功的全多尺度模拟，还有很长的路要走。建立鲁棒和快速收敛的数值求解方法需要进一步的研究。提出了进一步研究的需要。ASME版权所有©2010
Multiscale simulation is a rapidly evolving area of research that will have a great impact on computational mathematics and numerical modeling in engineering. In this keynote lecture following parts are included. First, what is multiscale problem. In the thermal and fluid science multiscale problems may be classified into two categories: multiscale process and multiscale system. By multiscale process we mean that the overall behavior is governed by processes occur at different length scales. By multiscale system we refer to a system that is characterized by a large variation in length scales. The cooling of an electronic system is such a typical multiscale system. Existing numerical methods for three geometric scales (macro, meso and micro) are briefly mentioned. In the second part the necessity of multiscale simulation is discussed. Examples are provided for multiscale process and multiscale system. In this lecture focus is put on the simulation of multiscale process. In the third section numerical approaches developed for the simulation of multiscale processes are presented. There are two types of simulation approaches. One is the usage of a general governing equation and solving the entire flow field involving a variation of several orders in characteristic geometric scale. The other is the so-called “solving regionally and coupling at the interfaces”. In this approach the processes at different length level is simulated by different numerical methods and then information is exchanged at the interfaces between different regions. The exchange of information should be conducted in a way that is physically meaningful, mathematically stable, and computationally efficient. The key point is the establishment of the reconstruction operator, which transforms the data of few variables of macroscopic computation to large amount of variables of microscale or mesoscale simulation. For different coupling cases the existing methods for such operators are briefly reviewed. In the fourth part, four numerical examples of multiscale simulation are presented: liquid flow in nanochannels with roughness by using MDS and FVM, flow and heat transfer in a micro nozzle by using DSMC in fluid and FVM in solid, flow past a cylinder and natural convection heat transfer in a square cavity by using coupled FVM and LBM. Finally, it is pointed out that we have a long way to go in order to have a successful full multiscale simulation for the complicated engineering problems as transport process in PEMFC and refrigerant condensation process on a enhanced surface. Further researches are highly required to establish robust and quick-convergent numerical solution approaches. Some further research needs are proposed.Copyright © 2010 by ASME