Towards Optimum Formulations of Large Eddy Simulaton of Turbulence

申请号：CTS- 9616219项目负责人：Moser这是一笔资助，旨在启动一个重点项目，开发可靠的LES方法，其准确性有明确的界限，可用于复杂的流体动力学应用。关于LES方程的详细推导，最优滤波器的定义和子网格模型的发展的几个基本问题将被解决。这项工作将侧重于在处理近壁区域和嵌入的小尺度层流子区域时需要遇到的挑战，这些挑战阻碍了LES在实际问题中的广泛适用性。在没有小尺度知识的情况下，大尺度进化本质上是随机的，这一事实的基础上，将使用一种新的LES形式来解决LES的突出问题。这种形式提供了一种技术，可以精确地测量大尺度演化的可预测性，以及任何特定LES模型的性能。此外，根据动力系统理论对LES的解释表明，过去的时间信息可以大大增加大尺度的可预测性。预计本研究将在LES的实践中取得若干进展。本研究的必要数据将从实验中的直接数值模拟（DNS）和/或粒子图像测速（PIV）测量中获得。这些技术将应用于三种特定的测试流动：平面通道、圆柱交叉流动和突然膨胀。DNS和PIV都是必要的，因为通过PIV实验可以获得更高的雷诺数，并且DNS可以获得更详细的三维数据。选择这三个测试流是因为它们是最简单的流，展示了在LES中造成困难的特性。在这些流的上下文中执行的优化和开发的模型将适用于各种流中的LES。这项工作将由NSF和AFOSR共同赞助。

ABSTRACT Proposal Number: CTS- 9616219 Principal Investigator: Moser This is a grant to initiate a focused project to develop reliable LES methods, with well defined bounds on their accuracy, that can be used in complex fluid dynamic applications. Several fundamental questions regarding the detailed derivation of the LES equations, the definition of optimal filters, and the development of subgrid models will be addressed. The work will focus on the challenges that need to be met in the treatment of near-wall regions and embedded small-scale laminar sub-regions that have hindered the wide spread applicability of LES in complex problems of practical interest. The outstanding issues of LES will be addressed using a new LES formalism based on the fact that without knowledge of the small scales, the large scale evolution is essentially stochastic. The formalism provides a technique to precisely measure the predictability of the evolution of the large scales, and the performance of any particular LES model. Furthermore, interpretation of LES in the light of dynamical systems theory, suggests that past time information can greatly increase the predictability of the large scales. It is expected that this research will lead to several advances in the practice of LES. The necessary data for this research will be obtained from direct numerical simulation (DNS) and/or particle image velocimetry (PIV) measurements in experiments. These techniques will be applied to three specific test flows: plane channel, cylinder in cross flow, and a sudden expansion. Both DNS and PIV are necessary because higher Reynolds numbers are accessible through PIV experiments and more detailed three-dimensional data are available from DNS. The three test flows were selected because they are the simplest flows that exhibit features that have caused difficulties in LES. The optimizations performed and the models developed in the context of these flows will be applicable to LES in a wide variety of flows. The work will be jointly sponsored by NSF and the AFOSR.